AVS vs. Milkdrop

Firstoff, this thread is not ment to be a "which one is better" thread. Everyone has there preferences. I'm merely starting it up because I have been designing AVS visuals for a while now and just recently found out one could design for MilkDrop as well. This makes me curious. What can be done there that cannot be done in AVS?

Hi solamon,

First off thank you starting this thread. Don't worry I do not intead turning this into a flame battle over what is better.

That said one of the main features in MilkDrop (MD) is the 3d acceleration and the speed at which it runs. This I feel is one of the things that lets AVS down.

It is difficult for me to comment in much detail about the diferences between them as I do not know fully what AVS can do.

For the most part MD is less configurable and can do potential less than AVS. In my opinion. From what I know about avs a main different is in the superscope and stuff that you have and MD does not. WE just have a collection of waveforms.

We have features that I do not think you have Borders, etc that are placed on the screen. The majority of the things are with th eplug-in as a whole, load time, preset blending. That I do not think AVS has?

Another note a couple of things surprise me about you (and maybe the rest of the AVS). That you only found out recently that MD was configurable (it is 6 months old from the first beta release) and that there is very little help for begineers out there to start AVS.

I wanted to have a go to see if I could add anything and write my own presets. I frequent the forums and the ******* room. Yet I do not know of any even a 1 page guide to of what the AVS features do. No help file? I understand the AvS can do soo much but it is many years old now and just thought that something useful will be available.

Do you know what AVS can do that MilkDrop cannot?

That is it for the moment. I have post a link to this debate on the MD forums and hopefully there will be plenty more chat about this. But the ball is rolling now.

CHeers,

Rovastar

While I would agree that there are quite a number of things that can be done in AVS, and not in MD, it comes at a price - SPEED.

As much as I like AVS, I can barely even run it fullscreen at 400 x 300 without slowdown, and MD can run on most computers at 800 x 600 (and potentially much higher) at 30 or 40 frames per second. This alone convinced me to use MD rather than AVS.

Other handy little features include having details about the song, being able to use 3D glasses, multimonitor capability, rating presets, blending, title animations, and a number of other options, which as far as I know, aren't in AVS.

I would urge all AVS users to at least give MD a try (go to the official MilkDrop page, and download more presets at Rovastar's MilkDrop page), and to see what you think. I think you'll be pleasantly surprised. I think you would find that many effects can be done with MD, and at a much better framerate. I would be interested to hear other AVS users' perspectives on this too......

I am also a MilkDropper, but I'm going to hold off my thoughts until some AVSers have had their say. I don't want to turn this into a one-sided thread ;]

- Krash

Hi there
Just downloaded Milkdrop, and while it seems quite pretty, it seems VERY VERY limited in terms of what it can do. Lots of fast moving blurred lines and thats about it. Now don't get me wrong, I am not trying to get into a flame war, but compared with the AVS versatility (such as avi's, bmp's, APE's), as a professional VJ tool, the AVS wins hands down over Milkdrop. Though it would be nice to have somewhere proper documentation on the AVS, say a page on each of the functions. Oh well we can dream can't we.....

it seems VERY VERY limited in terms of what it can do

hit www.milkdrop.co.uk, and download the user-made presets. There's a LOT more that we've been able to do with our simple oscilloscope that a bunch of fast-moving blurry lines =].

as a professional VJ tool, the AVS wins hands down over Milkdrop

I may be overracting somewhat, but I don't see how AVS could possibly be used as a professional VJ at anytime. On even simple presets, I have to reduce my resolution to 512x384x32 before I get a framerate that is consistently over 24fps. By contrast, I can run MilkDrop full screen at 1280x1024x32 with a 1024x1024 texture size and maximum mesh resolution at 30fps or so on all but the most complicated presets.

The simple factor of speed seems to make a difference to me. If I were in a night club doped up on acid, I'd much rather be watching some blurry lines that moved around at 60fps than a couple waveforms and some floating dots moving at 5fps. But maybe that's just me.

AVS versatility (such as avi's, bmp's, APE's

Agreed, these are some things AVS can do that MilkDrop cannot, but again, when it's running at 10fps, I don't really see the point.

I'd love for you to argue back at me about this. I don't want to get agressive or anything, but I'd really like to be shown the lights of AVS that everyone raves about.

- Krash

ok, i haven't tried milkdrop but i tired of the shit u MD:ers been saying. yeah of course avs is slow, we pick on it everyday, but is it nececery that u md:ers pick on it, u got yo fucking 3d shit that u do and we got our 2d shit we do, ok some avs:ers can do 3d but i can't. one more thing, i don't like the full screen shit, i think that fullscreen can be good if u like that.

so did i made my point clear?

Fair enuff with what you say as far as the user-presets go, I think that there's a lot that been done to push Milk-drop creatively.
As for using AVS as a professional VJ tool, well I use it all the time at parties and gigs across England, and it is done me pretty well. As for how I use the AVS I will post some presets up later today for you to look at and see what you think (they might be quite big as i will have to bundle the original AVI's I use with them 1-2megs). What it comes down to when using the AVS as a professional tool is that I can use it is a live filter effect unit on my own animations (and soon when the APE is done, live visual sources, such as video cameras as well). Now the idea is normally I use two machines doing this, alongside a couple of other video sources, and mix it all together on three serial linked vision mixers. As all these work on composite/tv signals only having a resolution of 352*288 is fine, the move to upgrade the hardware to digital resolutions is astronomical in the region of tens of thousands of UK pounds. Also as I am using the AVS to remix my own original 3d animations, it means that what you get to see, you ain't going to see anywhere else. Six hours of milkdrop would certainly bore the eyes, though I give you, it would be alright for very short sections here and there.
The other thing I have noticed is there is a lot of arguments over the fps, and pretty much from a biological point of view we can barely distinguish anything over 25fps, you normal european TV only works at 17.5fps, but cleverly makes it look like more with the use of interlacing two pictures to make it look smooth. So if I can get the filter running at between 15-20fps that does for most things. More would be nice, especially to do subliminal stuff, but that is the only real advantage of super fast fps (40fps+), other than the cheap thrill of tweaking numbers higher and higher.
And finally although I intend to look into Milkdrop more, for me AVS is also a lot more artistic, and intuitive to use. But then again it could be just me.
Looking forward to hearing all your opinions.

Now this sounds interesting... I'd very much like to see some of these AVIs as backgrounds, and other stuff you mentioned.

I still think anything under 30fps is noticeable, especially because whether AVS or MD, you're always going to have slowdown, and better to slow down from 30 - 20 fps than 12 - 8, or whatever. I can see that for your purposes, yodhe, that AVS would be better, but for general home use, I still side with MD. I hadn't thought of the fact that you could use it with vision mixers and so on.

Please post some of your stuff though, I would be very interested to take a look. I don't know about the size limits here though for posting external files, you may have to link the .avi's to your own webpage (if you have one). But I am looking forward to seeing this...

ok, i haven't tried milkdrop but i tired of the shit u MD:ers been saying. yeah of course avs is slow, we pick on it everyday, but is it nececery that u md:ers pick on it, u got yo fucking 3d shit that u do and we got our 2d shit we do, ok some avs:ers can do 3d but i can't. one more thing, i don't like the full screen shit, i think that fullscreen can be good if u like that.

so did i made my point clear?

I guess so, though we could probably do without the profanity. It didn't make your point any stronger.
And we weren't picking on AVS slowness. What we are supposed to be having here is a discussion of the relative strengths and weaknesses of the two vis plugins. And for the record, neither milkdrop nor AVS can do real 3d. It's all simulated (not sure about AVS on this one, correct me if I'm wrong). Part of the reason milkdrop is fast is that it utilises your 3d card to do a vast majority of it's transform operations. But that's not the only reason.
And if you don't like full screen, that's fine. But Milkdrop runs in a window as well, so it's up to a personal choice.

can use it is a live filter effect unit on my own animations (and soon when the APE is done, live visual sources, such as video cameras as well

Now that is pretty cool. Points for AVS there.

As all these work on composite/tv signals only having a resolution of 352*288 is fine

Valid point. Around here, our analogue tv signals are about 400*300, but it's the same argument. I'll add, however, that even if it is being filtered down to that resolution, you can still notice the difference in pixellation vs smoothness once it's been output to a tv (no comparison to AVS here, I've done it with milkdrop @ 400*300)

Six hours of milkdrop would certainly bore the eyes

The same could be said of six hours of AVS, even if it is putting music-based filters on an AVI. Unless you had a 6 hour-long AVI to filter, it would still get repetitive. Any filter based on music is going to repeat itself pretty often as well. You could say the same of filtering live feeds. Unless you have a live feed which is non-repetetive, it will get boring.

it would be alright for very short sections here and there

I don't know if you've actually looked at some of the presets we've put out recently, but there's a huge variety. I mean, speaking only of my own presets, I've got realistic-looking fire, Ocean waves, a sunset over the ocean, one thing which looks like a vinyl record, clock hands which jump around with the beat, a rainbow, a snowflake, as well as a variety of pulsating swirling colours in various (entirely different) configurations. Oh, and a physics-correct bouncing ball. And those are just a handful of my own presets. To be fair, there is an even wider variety of presets and effect available for AVS, though I partly attribute that to AVS's large user base.

from a biological point of view we can barely distinguish anything over 25fps... between 15-20fps that does for most things

Again, I have no argument there. I don't know the exact figures, but about 22fps is the generally quoted figure, I believe.
You have to consider though, that this 'biological fps' is for the detection of movement. Hence, blending effects look pretty much the same at 25fps as they do at 50fps. There is one difference that you CAN pick up though. MilkDrop (not sure about AVS, maybe it's configurable) redraws and blends the waveform every frame. It also performs calculations of bass, middle and treble levels every frame. This means two things: 1> At 50fps, the waveform is drawn twice for every 'frame' your eyes see. Your brain sees that this is happening, does some internal calculations, and makes you think that it's going faster. So even though you shouldn't biologically be able to, you can tell the difference between something moving at 25 and 50 fps.
2> because the music variables are calculated more often, they can also be more accurate or precise, and so you can get beat responses at high framerates that you wouldn't get at slower framerates. You can test this out with MilkDrop - it includes a framerate limiter.

me AVS is also a lot more artistic, and intuitive to use

It that respect, I'm very happy for you. I've taken a look at some AVS preset code, and I haven't been able to figure out what each of the variables are representing. In this regard, AVS could really use a help file.
The intuitive part could be that AVS is modular, whereas MilkDrop is not. If you want to do anything original or different in Milkdrop, you have to do it yourself. A good example is advanced beat detection. One of the more prolific preset authors (Unchained by name), spent about two months working on various beat detection algorithms, which got increasingly complex. The code for them is quite intimidating ;]

Please respond and continue the conversation. And any new arguments or ideas are more than welcome.

- Krash

Hi there
Here are the preset's i promised. They work on the avi files (purple ohm + agaric swirl) that can be found at www.psybrid.i12.com/videoloops.htm - remember to download them and place them in the avs directory or the presets won't work. These are just a few primitive examples of what I do from June 2001, and do accurately represent what I do but should give you an example.
Have fun
-Yodhe-

Argh

Sorry, just have to correct this one.

from a biological point of view we can barely distinguish anything over 25fps... between 15-20fps that does for most things

The human eye sees a framed animation < 25fps as individual frames. Any number above that will give a human viewer an impression of movement and animation. Note that if the animation moves slower, lower frame rates can still be acceptable and vice-versa, so this rule is not bound to one value.

However there is a difference between not distinguishing individual frames and not seeing a difference. Shake a pen around (not in front of your monitor). You should see 2 blurry pens at the extremes of your path, and a totally blurred/smeared 'pen' in between. The same applies for any object 🙂
Now do the same in front of your monitor (I'm talking CRT's here, for all your LCD/TFT buffs) You won't see a blurry path of a smeared pebn, but rather 7 or 8 individual copies of your pen.

What's the cause? Your monitor typically redraws itself 75 times per second (depends on your refresh-rate of course). This is done from top to bottom, left to right. So in fact, your monitor is a very fast strobe, which flickers 75 times per second. Yet our eye sees a constant surface, so we don't notice. However when we look at the monitor, we're not seeing a constant flow of light, but a collection of fast pulses.

So the difference is that in the first case, one 'frame' consists of all the light of the inbetween positions, smeared into one, blurry 'pen'. In the second case, the light of your monitor takes over, and you only see the frames when the monitor 'flashes'. So, while we don't *see* more than 30fps per second, we can clearly distinguish between animations that are 30fps per second or 60fps. In a graph, a 30fps animation is a 'staircase' model, where every step means a new frame. A 60-fps animation would be a finer-grained staircase that we would start to see as a constant line (there are 2 steps for each frame observed). For higher frame rates, this effect is even stronger.


Now back on topic for AVS vs. Milkdrop. Afaik, not many things in Milkdrop can't be done in AVS. The Superscope can be used to draw things such as borders, you can draw your own waveforms (3D superscopes) and even do advanced dynamic movements that simulate a 3D environment. AVS is also modular, completely customizable in the way the effects are ordered and easy to expand. Milkdrop trades this in for a huge speed increase.

Both have their advantages and I don't think you can say that either will 'bore' after 6 hours, given enough presets. The only thing that I do wish to say is that I'm not bothered by AVS's supposed 'low' resolution. I think most people view it in the wrong light: the resolution goes down, yes, but this is opposed by a much higher number of calculations and effects per pixel, which will always make the resulting image just as interesting as a double-sized hardware accelerated version. Low-res doesn't necessarily mean worse 🙂

Milkdrop/AVS Guide

I also think that an AVS guide is less needed than a Milkdrop guide... Milkdrop doesn't offer much inline help as far as I know, right? AVS has a list of all (well except for the 2 mysteriously hidden functions atan2 and exp) functions and their purpose, and usually contains a small text explaining the vars that one can use in an effect.

There's also a few tutorials around the net (and I've written some short guides here as replies to questions) but they just need to be bundled somewhere really.

Montana38,

Your comments disappoint me. You are slagging off a vis program you have never seen. For all you know it could run all the AVS visuals, have live feeds and make the tea at twice the speed of AVS.

It doesn't. It is different to AVS. But in summary it bridges the gap between old skool Geiss-like plugs and AVS. In terms of what it can do - preset writing, etc. Hopefully please most people out there. And the features are increasing all the time. I'm currently testing a new beta. And it again closes the 'gap' yet again I don't think you will be able to agrue against that. BTW Does AVS get regular and more importantly significant updates?

I do vis stuff but I am always looking for new stuff. To be honest I wanted to use my maths and MD knowledge to create some AVS. But there was no help for me.

In a year or 2's time there will more than likely something the blows away MilkDrop and AVS. I will want to change with the times and go for that - you should have more of an open mind.

To be honest what you said gives a bad impression on the rest of the AVS community. I know there are intelligent people on that frequent these boards.

This is not a 'AVS is really slow' thread there are enough of those already. But would you complain if the next release of the official AVS was 50% quicker. I think not.
========
Yodhe,

Interesting Comments. It does show again the poles apart in terms of what we like.

Ok back to features. From what I understand the beat recognition in AVS isn't all that good. Do you have variables like bass, mid_att, etc or is it just the BPM thing for guessing the beat.

MD to me is a lot more reactive to the music. If visuals are not reactive you may as well play music and watch the telly with the sound down or something.

Do you do VJ'ing full time in the UK? I am about to start VJ at weekends for mate who are DJ's/ promoters in various club nights in the Midlands mainly. But for the really big venues and one-off gigs. We may need extra special vis stuff. Maybe like what you use in addition to what MD has to offer. Could you drop me an email of how much you cost, hire for your services/ kits/ etc. Could be an additional earner for you.
rovastar@hotmail.com
==============
Unconed,

I was about to post and you have added more comments.

Thanks you for your comments. As always they were informative and interesting. Please respond to the beat recognation point I have made. Can AVS do this?

I understand the maths/speed ratio but remember MD can be slowed down also with really complication process and maths in the presets. The most basics AVS from what I have seen are slow.

And now you answered my AVS authoring guide question.

I would love to see what you would do in MilkDrop.
================
Krash,

Good points and Well done BTW well behaved 😉 and check Unconed AVS presets for the best examples of what it CAN do.

Phew,

Too much typing,

Rovastar

Excellent points gentelmen and nice work keeping things civilized. (Except for you Montana. 😠 What are you doing? Stop giving us a bad name. Why do you always have to stir up some garbage? Chill out, man. Damn.)

I have a couple of comments I'd like to post up. Much of it has been said already. (I wish I had posted up last night. Didn't think this thread would become this popular. 🙂)

About AVS needing a guide. I hear this from a lot of different people and everytime I whole-heartedly disagree. In my opinion, AVS does not need a guide. AVS is a program of infinate versitility. Any complete guide that was written would be probably be bigger than most of the Photoshop bibles out there. I liken learning AVS to being an apprentice wizard. At first he just has his basic components (the effects that come with AVS). Then he combines these components into basic spells (Render Moving Particle, Blitter Feedback) and these are added to his spellbook (basic combinations to make pieces of visuals). These spells continue to get more and more complicated the more he learns how different components work together. Eventually these basic spells will be able to be pieced together into more complex spells (taking MoveParticle and Blit, putting them together in an effect list, then putting a movement filter outside to warp the shape of the object) and with this his spellbook continues to grow. Soon the wizard is able to create entirely new compenents to form spells with (writing your own Superscope of Movement). Of course there is always going to be Archmages with gigantic spellbooks, but one can get along well with limited spells. This is not an attempt to brag, (obviously there are some visuals way better), but check out what I have been able to do and I have only been playing around with AVS since December.

But assuming such a guide were to exist, what would be put in it? The way I see it, all things could fall under the range of three topics. One, explaining how the program works on the interface level. This is not needed as AVS is very WYSIWYG and consists of very few buttons. Two, what each of the transformation and render effects do. Again, not needed. Just put it on the screen and see for yourself. AVS has the advantage of Real-Time visual creation. (Which, I'm uncertain if MilkDrop has presently. If not, it probably will soon.) One would certainly figure out what each of the effects do by playing around, at least as quick as a guide would teach. The only feature thats a bit tricky is the "effect list", which believe it or not, has a FAQ for it in the Develop section of the WinAmp page. Three, how to program superscopes and movements. Aside from explaining what each of the expressions does and explaining what the range of variables are that one has to work with (which it does, save for the two hidden expressions), what more could be said? Like UnConeD said a while ago, people want a FAQ to sum up in a couple minutes what must be learned over years in math courses. It's all mathmatics. You can't just teach someone that stuff so easily in a guide. When it comes to simple help-explain-this-to-me stuff, we got the forum and plenty of people willing to give a hand.

Another point I'd like to clairify, the whole AVS speed thing. This is my take on it. I have a P4 1.5gHz. For me, the best coded visuals (UnConeD, El-Vis, Tonic, etc) run at around 30-35 fps fullscreen at 800x600x32 (picel doubling on of course). What more do I need? I have no trouble with AVS running slow unless the preset has not been put together cleverly. AVS hinges on a good CPU. If one doesn't have it, then one must be content to run at a low resolution. Besides, not that I'm trying to reclassify AVS's slow side as a feature, but it is limitation that forces one to be inovative. What I'm about to say probably holds somewhat true for MilkDrop as well, but I think more so with AVS due to the stricter limitations. Creating a good visual is much like growing a Bonzai Tree. Careful pruning and clever planning create something very beautiful. However, if one lets every branch grow as it sees fit (ie: Visualization speaking: not being careful with what effects one uses) then the tree will come out looking lop-sided and lose it's ascetic appeal.

The last point I would like to bring up is; last night I downloaded some 700+ MilkDrop presets. This is not ment to sound critical but, it seems to me that there is much more similarity between one MilkDrop preset and the next than one AVS preset to the next. (of course I'm talking about talented AVS presets. Not the standard throw-a-bunch-of-effects-together preset.) I beleive that the main advantage AVS has is it's potential for infinate customiztion and creativity. If you don't have what you want, even down to the smallest detail... make it. Thats not to say that every MilkDrop visuals are the SAME as the next, just that AVS seems to have a much wider spectrum.

I think what would be needed right now is; if one of the MilkDrop people would be so kind, I would like for them to post up what are considered the best and most complicated MilkDrop visuals available. Then explain what makes it as powerful as it is. I would like to be shown some of the true potential MilkDrop development has to offer.

Thats about all I have to say. So, thanks for listening and I'm sorry to make this post so wordy. 🙂

Solamon,

As you are on the forums.

Quickly sorry it is speed again.

I don't want to make this a speed thread but you have just said
'This is my take on it. I have a P4 1.5gHz. For me, the best coded visuals (UnConeD, El-Vis, Tonic, etc) run at around 30-35 fps fullscreen at 800x600x32 (picel doubling on of course). What more do I need?'

Now I have 1.333Ghz with gf 2mx You also said it based on CPU speed. Not a bad machine I am sure ppl use worse out there.

I am running AVS now on my XP machine in desktop mode. A picked one of uncoded presets at random the Neon Shrooms one v. nice mind. Made the window exactly 800x600

I am getting 2.8-3.2 fps. WTF is going on then!

Did you make a typo?

Oh I am understanding more now.

pixel doubling on. is really halving the resolution in real terms you are running at 400*300*32!!!

Was that on purpose the confuse us MDers using special tech AVS terms? 😉
That's cheating🙂

Sorry about that! 🙂 I did not mean to confuse. Set Fullscreen mode at 800x600x32, vertical screenheigt to 90%, and pixeldoubling on. Don't resize the window, run it in fullscreen mode. AVS is very much so a program for people who like to tweak their computer to get the best results.

As for pixel doubling, you are not really halving the resolution. UnConeD, maybe you can field this one for me. I can't fully explain the technical specs on pixel doubling.

As for pixel doubling, you are not really halving the resolution

I've been watching it, and it seems to me that that is exactly what pixel doubling does. If you notice, with pixel doubling on, each individual square portion of the image is actually 4 screen pixels - two across, and two down.
If you don't believe me, make up a preset that has an oscilloscope with a very small amplitude. The smallest increment that the oscilloscope moves is one "pixel" of the image, but is infact 4 real pixels.
It's exactly the same as taking a 400x300 bmp, and blowing it up to 800x600 to sit on your desktop.

Put AVS on at 800x600 WITHOUT pixel doubling, running 100% of screen size, and you'll see the difference. The frame rate may be 1 or 2fps, but you'll notice that the actual image is smoother. Curves are rounder, slopes are less jaggy. What you're seeing is REAL 800x600.

AVS is very much so a program for people who like to tweak their computer to get the best results

I like to tweak my computer as much as the next guy. I fiddle with overclocking, driver updates, fiddling with hidden registry settings, etc. And it's always great if you can squeeze an extra 5 or 10 fps out of a game (or, in this case, a plugin) by tweaking things. But what is even better, is when you don't have to.

UnConeD - where can I find a collection of your best AVS presets?

- Krash

I still take that as 400x300. Sorry, less 10% for the gaps on the edge. 360x270 lol🙂. BTW AVS even calls it half resolution

Sorry couldn't resist the little joke there.😉

Ok attached is a zip some of the presets in MD that I do not believe AVS can do. And you can run them at a decent res. I suppose tweak the MD settings to get about 25-30 fps to get them right. On say Deeper Spacedust: Prob res 800x600 or higher if you can. tex siz 1024x1024, mess size 40x30.

These are some good presets maybe not the best but showing the different things that AVS cannot do (maybe).

Although to be far the actual res size is not true it is based on the texture and mesh size. Fancy Geiss stuff. It makes it near enough though.

Now from what I understand with AVS you have no good beat detection.
I have metioned this but have no reply to the post - saying that yours is such a busy forum. It has a mod ffs.😉🙂

These show some degree the beat detection available in MD. It is designed by us lots of the visual work for MD is behind the screens and not visual at all it is the beat detection. And we are still in the early stages of this.

Are you not impressed with this?

And as for the presets looking the same. We currently have only 3 active established preset authors (sorry other newer/less frequent posters but I was thinking of me, Krash, and Unchained) out there - only 7 people have post presets more than once. It is a small community. I think are really varied.

I would like to see the first few user written AVS presets 🙂

Just unzip these into a directory and play. Something like Deeper Spacedust try different music out with it to see the responses, esp breakbeaty types.
Most AVS I have seen do not react much at all to the music. Sometimes 'on the beat' but that different music being played through it seems to make little effect.

BTW when AVS is in full screen how can I make it jump to a certain preset say I want to look through all the Unconed ones?

Please drop your comments back.

Rovastar

Firstly, I want to answer a couple questions/comments that others have made, that haven't yet been addressed.

MilkDrop does have real-time preset editing. The hotkey to get to the editing menu is 'm', if you haven't found it yet.
And if you hadn't found that, then I would have to assume that you haven't read the readme, and so don't know the other hotkeys:
'l' brings up the load preset menu. You can browse to any directory to choose your presets from.
'p' brings up an internal version of the winamp playlist
't' will animate the song title
the F-keys toggle a number of different things. try them.

Also, there has been comments that AVS is a medium in which you have to have patience, in order to yield the maximum result (read: fps) from the limit that are imposed on you. At the same time, however, you make points that AVS has far more versatility, including things like AVI feeds, etc (which I am so far unimpressed with, despite the possibilities).
As a MilkDrop preset author, I have to deal with similar limitations. I have to generate interesting and new effects from limited tools (we don't have particles, or anything, just 8 different oscilloscopes of different sizes). This is where my limited tools are. What I have to start with is very limited in scope. At the same time, I also have versatility. Because with no fancy effects, and just a waveform blending into the background, I can get an easy 70+fps with all the settings cranked up at full screen. This ridiculously high frame rate allows me to add a large number of effects, including rotations, zoom, warp, combinations thereof, music responsiveness, colour changing, variations of all of these across the screen, etc.
But even after I add all these things, I'm still getting more than 30fps, because I started with a lot of exrta frames to play with.

Now, I'm on to the other purpose of this post. I'm going to run down a few of the presets that Rovastar posted, and explain what's special about them.

Digital Flame (myself) - a while back, we developed a way to generate solid, persistent lines of colour from the oscilloscopes (Rovastar's cosmic echoes, also included, is another example of this).
In Digital Flame, this effect has been combined with some musical responsiveness - the solid bars jump with response to the overall volume of the music. There is also a colour change - when treble is the dominating frequency, the colour shifts towards a light orange, and during bass, it shifts towards a pale blue. The majority of the time, the bars are a ruddy orange. when this is combined with a small-scale warp effect, and a cycle to prevent all areas of the screen from doing the same thing at the same time, we get very realistic looking fire.

OldSkool Blend (myself) - This was the first preset I did in which it had very intelligent colour cycling. The blending effects are very simplistic, as this was not my aim. Instead, I developed algorithms that will turn the waveform red during bass, and blue during treble, with an orange inbetween. I did this after repeated requests by users for someone to make a preset that looked like the original Geiss, with similar colour changes.

Ventilation and Mixed Disks (by Unchained and Rovastar respectively, based off a preset of mine) - These two presets utilise a similar effect, while using different methods of music responsiveness. Not only are these two presets responding in a different way, but they have individual methods of detecting when to respond, each coded by the respective author. Ventilation was an early attempt by Unchained to do more accurate beat detection, and he has since developed much more sophisticated and accurate methods. Mixed Disks is the first instance here of using a feature of MilkDrop - motion vectors - as part of the actual preset.

Accident in Paradise (Rovastar) - This preset is significant because Rovastar has worked around the problem of only having a single oscilloscope available to us. By shifting between the three states (in response to the music, no less, though in this instance, it's hard to tell), the illsuion of three concentric rings is formed. The only reason this illusion works is due to the high framerate, fooling the eyes into thinking the rings exist simultaneously.

Broken Phoenix (Rovastar) - This is a preset in which the musical responsiveness is very easy to see. Plus, it is the first example here of the Video Echo effect - in which the entire screen is drawn a second time. The alpha value, scale, and orientation of this second drawing can all be changed.

Cosmic Echoes (Rovastar) - Already mentioned to a degree, but with the additional feature of the appearance of two individual waveforms, generated by using the texture wrap feature.

Northern Lights Throbbing Bud (Rovastar) - A true fractal. Pulsating with the music. An easy 25fps. 'Nuff said.

Organic Rainbow (Rovastar, based on a preset by Bmelgren) - another variation on beat response, coupled with a simple example of some of the effects that can be generated with MilkDrop, using a minimum of code.

Deeper Spacedust (Rovastar, based on a preset by CTho) - a very easy to see display of the musical responsiveness achievable in MilkDrop. If you play the right music, you can see that the image stretches horizontally on bass beats, and vertically on treble.

Braindance Visions Remix (Unchained, based on a prest by Rovastar) - displays different effects occuring at different areas of the screen, with these areas being changed when features of the music reach critical levels. In particular, this preset sources it's values from the midrange as well as bass, which yields different forms of responsiveness to a simple "beat" detection.

Chess Brain Trip (Rovastar, based on a preset by Unchained) - A further display of our ability to break up the areas of the screen into different effects, and have these effects vary with the music.

Sea of Zigrot (Rovastar, based on a preset by Zylot) - A very simple showcase of music responsiveness. The colour brightens with midrange sounds, and the zoom increases with bass sounds, without necessarily responding to only the beat.

Imbibe and Take Your Pick (Studio Music) - Two more examples of the variety of music responsiveness, and the effects achievable with it. Take Your Pick also shows the tiling ability of MilkDrop, achievable through a Video Echo that has been scaled down.

Cyclic Insanity (Unchained) - Our first good example here of the border effects that MilkDrop is capable of, combined with a seemingly complex blending methods, and some of Unchained's earlier beat detection code.

Digital Warlord (Unchained) - Different effects in different areas of the screen, this time coupled with some of Unchained's most complex beat detection code.

Shadowbond (Unchained) - One of the pinnacles of Unchained's beat detection adventures, Shadowbond displays a variety of complex effects which jump seamlessly from one to another in almost perfect response to the music.

Digital Energy (Rovastar, based on Digital Warlord) - More variety in how the preset responds to the beat, as well as Motion Vectors to add further interest.

For The Seagull (Unchained) - The beat detection isn't up there with Shadowbond, but it is still effective, only this time the overall effect is much more soothing.

Well, I think that about covers it. There are variations on all of these, and even more styles besides, but these showcase much of the variety that MilkDrop is capable of.

- Krash

I knew Milkdrop had realtime preset editing, but I've looked around in it and it felt very sluggish. Having the menu on top of the visual is neat, but it also limits the interface to keyboard only. Although you probably get used to it in no time 🙂 On the other hand, AVS has no interface whatsoever in full-screen mode, and you can't define a playlist of presets, or even choose what you want. And on top of that the random feature (press space) sucks large paper balls sideways through long straws. It often shows the same preset.

As far as movement is concerned, I believe MD has a big advantage over AVS here... AVS requires you to define the relationship between source and target coordinates (either [x,y] or [d,r] (polar)) without any extras. Milkdrop has rotation, zooming, warping, etc. all built-in, for easy access. No to mention the different dx, cx, sx, etc. What do they do? (relative movement, movement of the center, ...?). In any case, in Milkdrop, this will all be combined into one big movement 'together' (one final grid shape). In AVS however, if you wish to use the built-in effects to get both a rotation, shifting and zooming, you'll get three phases of movement. If you know the math behind them, you can just as easily combine them into one, but not many people do.

I'd also like to point out the immense complexity of most blend modes. Do them on a CPU, and you'll take several valuable cycles per pixel, even when using MMX and newer CPU ops (not sure if SSE is useful for pixel ops). Do them on a chip that's specifically designed for it, and you can literally do them for free since you can lay out a circuit that does alpha-blending in a whip.

As far as pixel-doubling is concerned, I didn't think this would stir up any confusion really. It just doubles the output size: internally everything is at the original res. Useful for windowed mode, but not fullscreen (unless your video card can't do small resolutions).

Now to talk about AVS's scope of effects. The problem is that some have been around for a long time, while the newer (programmable) ones are not. So some effects (e.g. clear screen) are not as customizable as you'd want them to be, so you need to create a similar effect using replacements. AVS's beat detection is not fabulous either. You just a 'is there a beat?' value. However you can also grab any value on the waveform or spectrum using getspec or getosc with a variable width. So you can grab 'all the bass' or 'all the treble' too. Beat-detection is just something that's not focused much on in AVS.

While AVS also has plug-in effects, I'm not satisfied with them. I've written a few test APEs, and some main problems are:
- No access to any AVS global variables (blend mode, line width, etc)
- No access to extra AVS buffers
- No way to use the nifty scripting language

I'm actually writing my own AVS (and MD) compatible compiler to fix that third issue, but I'm just re-inventing the wheel here.

Maybe either dev team needs to look at the 'competition' (they're both Nullsoft 🙂 and copy some things... I'm sure it would result in a better AVS *and* Milkdrop.


For my presets, either look for "Whacko" on this site, or visit http://avs.acko.net/ which is the AVS section of my website.

UnConeD,

If MD is not changing presets mate then something is wrong.
I admit the sometimes the 'randomness' is always spot on.

Press L on the keyboard to manually change them.

If it is sluggish use the slow text fix option maybe.

MD redraws things every frame first working out what every pixel is doing. SO I don't really understand what you mean with the 3 phases of movement. We can have many different effects doing off at once? Maybe I am not tech enough to understand.
And nearly all the effects in MD are programmed are you ever looking at the per_pixel and per_frame sections of the code?

Actually the zoom, rot, etc are all complex equations from the dx and dy controls.

dx - controls amount of constant horizontal motion; -0.01 = move left 1% per frame, 0=none, 0.01 = move right 1%
dy - controls amount of constant vertical motion; -0.01 = move up 1% per frame, 0=none, 0.01 = move down 1%

For the explaination of what all the variables, sy, etc do is here:



You can actually get to it from the plug-in config box. Yep we have one🙂

another begineers guide but explained in more detail on my website: www.milkdrop.co.uk


You sated in a previous post.

not many things in Milkdrop can't be done in AVS.

What about all those presets in the last zip?

Do you stand by that? I would love to see something like Deeper Spacedust emulated on AVS. To compare. That is if AVS can do everything MD can.

What does everyone else think???? Come on people there are 1000’s upon 1000’s of AVS writers out there you must have some opinion.
Has anyone actually tried MilkDrop?

Remember this is only the tip of the iceberg you AVS have been at it for years. We are newbies to it. They are getting better all the time the more we know.

And the reason that fps is important is that the beat responsiveness is directly proportional to how good the preset works. Since the thread started I have been looking at AVS in details and most (nearly all) of them have no beat recognition. Now for me a visual plug-in based on the music is a fundamental part - Otherwise it is like looks at a load of crude avi's. If that is not being too harsh.

Has this feature been requested! It needs to be having variable for actual and attenuated sound for the bass, treble and mid range in a sscale.

==============
My summary of AVS and MD is that AVS can more than MD creatively wise but if MD can do it then the effect (speed/detail (res)) is about twice as good. And AVS cannot really do beat response stuff. And MD can do a hell of a lot!! More I think than you realise.
==================

Come on people comment more on this thread. I have seen many AVS on these boards over the last couple of days and we only have had a couple of downloads if these presets.

===================
Do you not like MilkDrop because you lot cannot do the maths and need your little GUI's is that why you use AVS??
====================

(Hopefully that cheeky and maybe sarcastic comment should get some posts)

Another huge posting.

Later,

Rovastar

Um?

If MD is not changing presets mate then something is wrong.
I admit the sometimes the 'randomness' is always spot on. Press L on the keyboard to manually change them. If it is sluggish use the slow text fix option maybe.

I think you misread my post. I was talking about AVS there. And with MD's 'sluggish' interface I think I should've said 'hard to use'. I'm not a user who wants to do everything with a click (I use the Command Prompt a lot in Windows just because it does certains things waaaaaay better), but AVS's GUI clearly scores some better marks there over MD's keyboard-only interface.

Has this feature been requested! It needs to be having variable for actual and attenuated sound for the bass, treble and mid range in a sscale.

I'm not sure what you mean with 'attenuated'. You can get a general bass level through getspec(0,0.1,0) <- this will grab the spectrum data around point 0 with a width of 0.1 and for the center channel (0). If you want a general VU meter, use a very wide range. If you want treble, center around a higher value (e.g. 0.7). If you want an exact value for one spectrumpoint, use width 0. You can grab any part of the spectrum like this really, both exact values or averaged regions.

Do you not like MilkDrop because you lot cannot do the maths and need your little GUI's is that why you use AVS??

Please explain how a GUI with nice editboxes and lists can get around the fact that programming your own effects requires math-skills. Sure, we have nice sliders for zooming and all, but to get anything decent done you need to move away from most of the built-in effects anyhow and use your own dynamic movements or superscopes.
On top of this, I think I can show that AVS is *harder* to program than MD:
MD has variables cx, cy for the center of rotation/strectching, you have variables sx, sy to do stretching and variables dx, dy for regular shifts.
In AVS, you can only define the relationship between either a cartesian coordinate system (x,y) and a new (x',y') or a polar coordinate system (d,r) and (d',r').
Let's go back to this:

MD redraws things every frame first working out what every pixel is doing. SO I don't really understand what you mean with the 3 phases of movement. We can have many different effects doing off at once? Maybe I am not tech enough to understand. And nearly all the effects in MD are programmed are you ever looking at the per_pixel and per_frame sections of the code?

When you want to do a shifting, rotating and stretching effect in Milkdrop, you'll just set the three sets of variables accordingly. Then, Milkdrop will combine them by calculating *one* final grid-shape that defines the global shape of a frame in relation to the previous one.
In AVS, when you use the built-in effects serially, it'll first shift, then rotate, then stretch, running over the image three times (slow!). To avoid this, you need to write your own 'dynamic movement' (the AVS equivalent of what Milkdrop uses for movement -> a grid) and write out equations for rotation and stretching around an arbitrary center using conversions between polar and rectangular coordinates (cos/sin in one direction, atan2 and pythagoras in the other direction). This will result in the dynamic movement grid taking on the same shape as Milkdrop would have it.

The difference is that in Milkdrop, you can only have one global grid-size. In AVS, you can define any number of transformations in any order, but it'll run over the image several times (i.e. perform each operation separately), which is why it'll be a low slower usually.

UnConeD,

Quickly I was being sarcastic🙄 🙄 🙄 🙄 🙄 🙄 🙄 🙄 🙄 🙄

about the maths thing. I know what you do in AVS. I know you know the maths😉 I know AVS can do more I was trying to get a response.😉
You are shitloads more tech than I am!!!

But I don't think you know how much maths/programming does into some of these presets look at Unchained or some of Krashes ones or my boxfresh ones. Please download and look. So we can talk about the same thing. The per_frame and per_pixel sections are huge. Most of it is not even visual it is beat detection etc.

attenuated -

eg bass_att - retrieves an attenuated reading on the bass, meaning that it is damped in time and doesn't change so rapidly.
as in the MD docs. In the link before. Read for more info it is different to the bass variable.


The rest of the stuff is too tech for me to comment. But thanks for the input anyway.

What I was looking proward to was a reply to this:

You sated in a previous post.

not many things in Milkdrop can't be done in AVS.

What about all those presets in the last zip?

Do you stand by that? I would love to see something like Deeper Spacedust emulated on AVS. To compare. That is if AVS can do everything MD can.

Later,

Rova

Well actually the attenuated bass can be calculated like this:

bass_att = bass_att * .9 + getspec(0,0.1,0)*.1

In both Milkdrop and AVS. Of course I don't know what exact formula Milkdrop uses, but it should be close. And I did in fact download the presets and view them 🙂. They contain neat code for beat detection, but the visual side is only so-so in my opinion.


I decided to take up on your challenge and port Deeper Spacedust to AVS. Because I don't know what scale the attenuated levels are in Milkdrop, I improvised a bit. I also used AVS's built-in beat-detector instead of the code in the preset. And the colors don't match, but that's because they depend on the (unknown) attenuated levels.



Comes pretty close, but it's different.

Do I need to download something else to view that preset? All I see is a white spectrum analyser, on a black background, with no fading. I got similar problems with a pack your other presets that I downloaded.

Oh, and FYI - the bass_att values aren't what you've thought they are. Maybe Rova led you to believe wrongly, I don't know. Here is the basic equation:

Bass_att = bass(this frame) / bass(previous frame)

It's a relative percentage thing. There might be more to it, but that's the gist of it.

Although you probably get used to it in no time

Aye, that you do.
For most presets that I write these days, I don't even touch the actual menu part, except to change the waveform. I usually do everything in code.
Once you're used to it, the on-screen interface is quite useful. It's cool to be able to be sitting watching it go full screen, and then a preset comes up that you realise has a mistake, or would look better in green, or whatever - you can change all that without ever seeing your desktop.

And on top of that the random feature (press space) sucks large paper balls sideways through long straws.

Graphic description there. 😉 👍

I understand what you're talking about with the difference in movement, even if Rovastar doesn't =]
As you've figured out, all the calculations in MilkDrop are done inline, before Anything is changed on screen. So MD grabs all the values that are set in the menus, then runs through the per-frame code. After that, it runs through the per-pixel code once for every grid point on the mesh. Values for all the remaining pixels are taken as a weighted average of the mesh points around them. Only after all this is done, does MilkDrop draw the next frame.

As you've said, AVS is different, in that the screen is recalculated for every module in the preset. And yes, this is part of the reason AVS is slow, along with the fact that it does everything on the CPU. SSE / 3DNow! might be useful for pixel ops, these are basically streamlined loating-point operations. So if you're moving things around using decimals, they might come in handy. (They're referred to as 3d-accelerating instruction sets, because when simulating 3d space, almost everything is a decimal)

Beat-detection is just something that's not focused much on in AVS.

I agree with Rovastar on this one. I admit there are many MD presets (most notably the ones it comes with, but many user-made ones as well) that have no reactivity to the beat at all, beyond the oscilloscope. At the same time though, we have many which do react noticably to the music, and we're developing new effects all the time. What's the point of a music visualisation plugin that has no discernible relation to the music? =P
The presets that don't react with music (in either MD or AVS) could just as easily be AVIs, and you'd never know. Plus, they'd be pre-rendered, so they'd run faster. 😉

I was previously unaware of AVS's ability to "grab" info from a point (or more usefully, a range) on the spectrum. This then become your equivalent to our bass, mid, and treb variables. while being able to arbitrarily select a point on the waveform is cool, I don't see what you'd do with it. I would probably be constantly defining the same areas of the spectrum to refer to as bass, or whatever.
Defining a single point could posibly be used as a method for finding a random number, but why bother, when a random function already exists?

I think I can show that AVS is *harder* to program than MD

could you give us a simple example? What would you need to do if you wanted to move pixels further to the right the further from the left they got? What I mean, is, have things moving very slowly at the left side of the screen, and then accelerating exponentially (or even linear acceleration) off to the right. Would something simple like that be alot of work to code in AVS?

The superscope is one part of AVS that I haven't been able to figure out. I've been looking at some of El-Vis' 3d presets (maybe they're really complicated, but anyway). He seems to be using alot of variables in his definition of the superscope. Things like rx, ry, and rz (which I am assuming are for rotation along the three axes). What I'm confused about, is which variables are part of the actual superscope, and which were defined by El-vis for the purpose of his preset? An of those which ARE part of the superscope, what do they each control?

I think that if I understood AVS a little better, I may be more appreciative of it. I don't think I'll ever prefer it over MilkDrop, as the speed more than anything is such a big issue for me. But I'd like to be shown just how versatile it is, by some preset writers who have done supremely excellent presets.

I like this conversation. Let's continue it.

- Krash

Actually the superscope is a very simple effect basically. You just set an (x,y) value for each point on the scope (and a red,green,blue value for color at that point) in relation to 'i' (which is a percentage of which point we're at... e.g. with 20 point superscope, i will move in 1/20th increments).

So to do a 3D scope, you need to calculate everything yourself. All AVS looks at is that (x,y) pair 🙂

Woohoo it seems my explanation of the movements was not totally unintelligble. Sorry, sometimes I suck at explaining things. The problem with AVS is that you can either do (x,y) coordinates or (d,r). But not both of them at the same time.

So if you were to do (in MD):
cx=-1; cy=-1; rot=0.01;
(I mean a rotation around the top-left of the screen, not sure of the 'rot' usage).

You'd have to do this in AVS (with a dynamic movement):
(switch rectangular coordinates on)

x=x-(-1);y=y-(-1);         // move the origin to (-1, -1)
d=sqrt(sqr(x)+sqr(y));     // calculate new (d,r) coordinates
r=atan2(x,-y);
r=r+0.01;                  // increase rotation
x=sin(r)*d+1;y=-cos(r)*d+1;// go back to (x,y) pair and translate origin back 

Not exactly easy.

By the way, getspec's second parameter defines the width of the area to grab. A width of 0 will give you an arbitrary point on the spectrum, but if you use getspec(0,0.2,0) for example, you'll get all the bass averaged (grab all values around '0' in an area of 0.2 wide). So you can indeed use them as replacements for bass/mid/treble. Thanks for clearing up the bass_att thing too by the way. I'd call it the derivative of the bass, or the rate of change 🙂.

Defining a single point could posibly be used as a method for finding a random number, but why bother, when a random function already exists?

Well actually it's very useful to use it as a replacement for rand(): it's always the same across one frame. So if you want several effects to behave randomly, but have the same series of random inputs, you use getspec. An alternative is to write your own random number generator which I did in a few presets ('The Source' in Whacko AVS II) when you need yet another property (the same series of random numbers across all frames).

All points taken at hand but < Seeing is believeling > and Milkdrop blows AVS out of the water. SM

That's your opinion, you make it sound as a general truth. Everyone has different views of what he/she considers to be important so some will find AVS better than MD, or vice-versa.

I could just as well post a topic "Seeing is believing: AVS blows MD away because it offers more flexibility, plug-in effects and a better GUI."

I thought we were not going to flame eachother on which is best, but rather try to compare them and learn new things from one another?