V1.03 beta bugs

[Rovastar]

Ok post here with any bugs that you have with v1.03 beta.

Here is the list so far.

-For the normal text options ~11-15 'testing' the bottom of the g is always missing.

-wave_a is the opacity that is fine but for some reason it des not work for wave 3. Work for all the rest but 3?

-wave_type does nothing at all!

-q variable do not work in the init section.

-motion vectors messed up (in detail below)


From Krash
------------
The motion vector things are a bit b0rked now - it's gonna require recoding all the vectrip presets (basically any preset with moving mv's... sigh)

I've been looking briefly at the way the mv's work now - it seem like Ryan tried to implement my suggestion, only it doesn't work right. Needs some more work on it to get things fixed.

ATM:
mv_x = 1 - one motion vector, right side of screen
mv_x = 1.5 - in center of screen. So far, it's like I described.
mv_x = 1.79 - that single motion vector continues moving past the centre...
mv_x = 1.8 - all of a sudden, a second motion vector appears. strange value to have the secnd one appear.
mv_x = 2 - the two mvs are now equally spaced around the centre. By my description, this shouldn't have happened until 2.5...
mv_x = 2 - 2.99 - the vectors move around a bit...
mv_x = 3 - the third vector appears, but the vectors aren't centered on the screen...
mv_x = 3.5 - now the vectors are centered.
mv_x = 3.8 - the fourth vector appears
mv_x = 4 - the four vectors are now centered.

This seems like a weird system. The setup that I described was designed to always have the motion vectors centered on the screen (except when there was only one). For example, and value of three would have one vector in the center, and one on each side, 3.5 would have them equally spaced, and as the value approached four, the vectors would come closer into the center of the screen, until at 3.999, they gave the appearance of a single vector. As soo as you hit 4, two more vectors would appear on the outside, and the three vectors that were in the center would now be two vectors in the center. increasing the value would see the side vectors move inward, and the center vectors move ******d. at 4.5 the vectors would be evenly spaced again. get to 5, and you have the four vectors having converged to two, and three new vectors appearing - one on each side, one in the centre.

It's a complicated system, and I'd hate to have to code it. If this is what Ryan was trying to do - much respect... I wouldn't want to try...

I suppose that now, we have to go and recode the moving vector presets to use the scrolling speed... except that the scrolling speed doesn't seem to work right, and I can't figure out a variable name for it (tried variatious combinatons of mv_sx... no dice.

----

Motion vector scrolling speed seems a little weird - set it to a constant value, and the mv's move, slow down a bit, and then suddenly speed up again. It's not in a sinusoidal variation of the speed... kinda strange.
Actually, the scrolling speed seems a little warped altogether. Set the vertical scrolling speed to 5, and you get the effect I described. Then, if you jump it down to 1, the mv's scroll *the other way*... might want to omit this feature if you can't get it to work.

motion vectors are drawn even though the motion is zero. It would be nice if we could turn this off (I think I could do fireworks if I could turn this off)

Images: works nicely - I can't find how to turn the images off manually (i.e if they don't have 'done' code in them) I won't have too much use for this myself, as I don't do any VJing, but it looks pretty useful for those that do.

One problem - in some blendmodes, brightly coloured images make things in the background visible that otherwise appear black - this is IMO, down to a fundamental problem with the way decay works. Whenever a part of the image gets to a certain brightness, it should just be set to x000000 (black). Otherwise, you can get funny visiual artifacts with the images. Example: I was testing the images with my "blank" preset - wave 0, decay 0.9. When I put an image over the top which is mostly white, I can see the song titles and text messages that (I thought) had faded to black.
------------------------

OK if you have any more add them to this list and hopefully it will be fully released in the next week or so.

Rovastar

[Illusion]

OK.... here's my thoughts... some bugs & suggestions.......

1) For some reason, "fixing" the dx/dy problem has actually made it worse, at least on my card. The test presets (in the various texture sizes, 256 x 256 and so on) all seem to "wobble" now, whereas previously it would only occur in 2048 x 2048. May just be my dodgy vid card however..

2) Agree with Krash & Rova on the MVs & scrolling speed

3) Same problem as Rova with #11 - #15 messages

4) Is wave_a accessible from within MD? Maybe I missed it, but I thought it would be called wave_a? Or is this just opacity under the waveform menu?

5) Images, like Krash said, need an "instant kill", rather than hardcoding the time the image is on the screen. Perhaps use another key for this?

6) Need some way of knowing whether we can input messages OR sprites. Perhaps a message similar to the one which appears for a few seconds when you choose random or sequential order? Or use another F key to toggle the mode to sprites, messages or off??

7) With the sprites, I assumed that repeatx or repeaty would mean it would display the image tiled, such as if repeatx = 6 and repeaty = 6, you would get 36 images on the screen. This seems to go wrong on my machine though, only one instance of the image displays correctly, while the others look all messed up.

8) More options for the sprites, if possible? Perhaps a zooming effect, so that you could slowly fade away a picture or logo into the distance.

9) How does size work? I have tried using it in the init_1 or code_1 lines, ie.

init_1 = size = 0.5;
OR
code_3 = size = 0.5,

but neither of these seem to work. Also, what does this mean: "size of 1 means the [largest axis of the] image will take up 100% of the
smallest edge of the screen."

10) Instancing.... loading the same texture seems to use a lot more video memory. For example, pressing "5" six times completely kills the speed, whereas I got the impression it would be accessing the same image, and therefore not require as much video memory.

11) This is probably coming in the proper version, but more documentation on how to use images. The examples are good, but it took me quite a while to get the hang of how to use the sprites.

12) Possibly being able to cycle through multiple jpgs under one number? Rather than pressing 01, 02, 03, you could maybe say img = C:/1.jpg; C:/2.jpg; C:/3.jpg, and the effect will go through each of these in order, kind of like a slideshow. This probably isn't possible though, but would be very cool :)

13) Drop-down boxes (ie. select image, effect etc. would be really handly, but again, not sure if this would take too long to do...

OK, now some general comments. I really like the new sprite idea, there is a lot of potential there. I had assumed that a background would be part of a preset, but it works much better this way. A lot of the presets look a lot better in 16-bit colour, which is good. The other new options (thicker waves etc.) are also excellent.

Well, I think that was it :) Any one got any comments or more suggestions / bugs?

[Rovastar]

OK I have not studied the images in too much details so cannot comment too much.

1) Eek! I don't understand this. In v1.02 if you had changed teh dx/dy offset by hand did this occur. Using the actual offset of -1/2*texture size. Rather than the guess? of 0.0005, etc. I do not know what to do about this.

You card probably cannot do that much detail of fine equations or something.

4) Wave_A does work but not for wave 3. Remember modulate opacity by zoom needs to be turned off obviously for this to work.

5) already suggest to Ryan '-##' for deleting the first occurance of that said image. I also want 'blend away' a la text messages.

6) It would be nice as an option. But why 2 keys for text and images. Why not one say 'Y' for toggling images/text like 'R' toggles random/sequential

10) Illusion first off remember you are using an old video card.:);) There is a little slowdown but not as much as a new image being loaded. It is probably significant on your PC.

11) I have not done much on this at all so I cannot but if you Illusion know something then why don't YOU write a more detailed guide on this. It could be included in the standard documentation if not included thee it will be on milkdrop.co.uk. Then Ryan focus on these bugs/ and/or more features.:)

[Illusion]

1) Yeah, it's probably just my video card, it doesn't really bother me anyway, so don't worry on account of me, unless anyone else has some problems... I was using your test presets when we were figuring out the problem, in v1.02 they all stayed still apart from 2048 x 2048 (I assumed the offsets were part of your code in these test ones?). However, the presets at all texture sizes move just slightly in v1.03, it's not so bad really

4) Maybe I missed something, but wave_a IS opacity isn't it, in MD??

5) Excellent idea, that way you know exactly which ones you're using, and which ones you're not.

6) I like that idea too, hadn't thought of that. Would one of them (sprites / messages) need to be set as default when you start MD, or would it remember from last time?

7) Are you having the same problem? Have I misunderstood what repeatx and repeaty do? I've got a screenshot here of what it looks like using these... repeatx & repeaty are both 4. There should be, I assumed, 16 pictures of the car, but instead they're all messed up. I had to decrease colours and size to get the attachment small enough to fit though. Try it with your own texture... just change the "img =" line.

Here's the code for it:

[img00]
img=d:\a1.jpg
colorkey_lo=0x000000
colorkey_hi=0x808080
init_1=blendmode = 2;
code_1=rot = time*0.27;
code_2=new_scale = 0.75 + 0.1*sin(time*0.6);
code_3=sx = new_scale;
code_4=sy = new_scale;
code_5=done=above(frame,400);

8) and 12) Are either of these even possible to code?

9) If anyone knows how to use size, please let me know :)

10) Yes, I know my video card is slow, don't keep rubbing it in. ;) I just got the impression it would use no more video memory at all from what Ryan said - something like "loading multiple instances should not use any more of the video card" or something. BTW, does putting one texture on give you much of a framerate hit? It's not too bad on mine really... drops from 30 to maybe 25 - 26...

On that subject, loading textures in MD has had a few unusual effects in Win98 when I exit, like icons not displaying due to lack of memory and so on - but I suspect it is just Windows needing yet another reinstall :mad: It's doing all sorts of weird things.

11) That's the problem, there's a lot of stuff I can't make sense of yet (size, repeat x and so on) so I don't think I'd be qualified to write a guide. Well, not yet anyway :)

13) I'm sure this would take ages, so never mind.

----------------------------------------
Finally, maybe people (if they find a really good effect for the sprites) could post their code & pic for the milk_img file? You could put as a zip the few lines that are applicable in a text file, plus the texture that you used? These could be posted to the preset forums maybe... just a thought...

[Krash]

From an email I sent to Ryan:

I had an idea for what you could do for the mv's in 1.03, to allow us some more freedom, but still have things backwards-compatible.

Leave the calculation of mv_x and mv_y the same as previous versions, and add two new variables, which would be used an an offset variable.

So if you had an mv x offset (call the variable mv_ox, maybe?) that was set to -0.5, all the motion vectors would be shifted half a screen to the left. A value of 1 would shift all the vectors an entire scree to the right (so they would all be off screen). Possibly have an additional yes/no variable to decide whether or not the vectors would "wrap", and come back on the other side.

This way, if the author desired it, we could program the motion vectors to move (by changing the offset). It would also be possible to have a couple motion vectors in one corner, and nowhere else, or a row of motion vectors along the top of the screen, etc.

/email

Also, what does this mean: "size of 1 means the [largest axis of the] image will take up 100% of the
smallest edge of the screen."

say for example, that you're running in 800x600, and you load up an image which is 50x100. have size = 1 means that the largest axis of the image (100) will be scaled up to equal the smallest axis of the screen (600). So the image will be scaled up to be 300x600. I imagine that size = 0.5 would make the image 150x300, and 1.5 would make is 450x900. Basically, a size of 1 will mean that the image is as big as it can be without some of it going off the edges of the screen.

Oh, and having 5 images running at the same time is more likely to affect the fillrate of your card than its memory - if the images have a source-based blendmode (you can see through them partially, or something), and they're reasonably large, then milkdrop basically needs to be redrawing the screen once for every sprite, plus once for the actual preset. It adds up.

- Krash

[Illusion]

OK, I understand what size means now. But, I still can't get it to work.I tried both of these lines:

init_1=size=0.5;
code_2=size=0.5;

First, by themselves, then together, but the picture stayed the same size in all cases.. any ideas?

I like the idea with the MVs, makes more sense than before.

Out of curiousity, has anyone tried this yet: Ryan's new plugin (http://www.geisswerks.com/monkey_beta2.zip), it doesn't work on mine (though it's almost definitely that my vid card just can't handle it).... looks cool though.

[sysRq]

hmm... monkey is strange... what exactly is it? the beginning of a new screensaver ryan is working on or what? it runs very fast and looks very pretty, imo better than smoke by far... is it meant to go with music in any way or is it just a continual tunnel?

[Rovastar]

From beta 1

-----------
Hey all,

Check out this new viz based on metaball cross sections. It looks pretty nice, but it's very nascent - I just got the 2D metaball cross-section-connection code working this afternoon, then I slapped it into a 3d environment and came up with this. It looks kinda like in the end of 2001, but in 3D.

http://www.geisswerks.com/monkey_beta.zip

IMPORTANT: unzip it to C:\Z. If you don't do this, it won't be able to find the textures, and the whole thing will appear black. Yes, it is lame.

Requires DirectX 8 (which, by the way, I touched for the first time today, and am instantly in love with.)

No audio hookup yet, but that will come soon (this will turn into a plug-in).

Hope you all like...

Cheers-
Ryan
-------------------

[ryan]

Beta 2 bug.

The "seceret" waveform is disabled like in the per_frame say you have...


wave_mode=8;

It would start as wave 8 for about 5 frames then go to wave 0 but its offcenter, more to the left.

Try it on your computer and tell me if it does it too. If so I have found a bug.