[Vasudan Admiral]

I definitely am.
However, considering I'm not any good with maths yet - this kinda thing is gonna take a fair while to become reasonably decent with.

Also, remember that there are a lot of other more pressing needs than making weapon effects for campaigns. I expect people will learn to use it when the they decide it's time to do some, but until then they'd be more interested in getting fredding, modeling and texturing done.
Just be patient I guess.

Short tutorials would be a very good and helpful way of getting more users too BTW.

[Bobboau]

alrighty then, the zip should be updated, I fixed the sublayer evaulateing too often bug and I'm fairly sure I sealed a rather nasty memory leak, I also did a bit of work on getting the progress of layer evaluations displayed a bit nicer (and you'l note there is a status bar on the bottom that gives you some information on what's going on), and I fixed a bug related to loading layers that reference each other out of sequence (although it still has the potential for problems). I'm half dead sleepy right now so I wouldn't rule out me doing something stupid that would cause a crash, but it should be working better than ever now.

I sort of have on the back burner of my brain this idea of adding in image import and editing layers and vector layers, that would make this more powerful than photoshop in a raw sense.

[Vasudan Admiral]

Oooh, the import image idea hadn't occured to me, but on a number of occasions I wondered about being able to preserve a layer as an image rather than code - for example if you happened upon a perlin noise layer that worked brilliantly for what you wanted.

Saving the noise layer and then re-importing it as an image to replace the code version of the same layer would be quite handy, and would mean one less evaluation you'd have to do when re-loading a project.

Anyways, seems to work fine for me - no crashes.

[Water]

I fixed a bug related to loading layers that reference each other out of sequence (although it still has the potential for problems). I'm half dead sleepy right now so I wouldn't rule out me doing something stupid that would cause a crash, but it should be working better than ever now.

I sort of have on the back burner of my brain this idea of adding in image import and editing layers and vector layers, that would make this more powerful than photoshop in a raw sense.

Loading layers is a bit hit and miss (crash) It's almost random

loading a new layer can overwrite the code of an existing layer

This version does seem a bit more unstable.

The other ideas sound good. Image import would be great. Vector layers sounds interesting. How much work on your part does that involve?

[Bobboau]

a lot, especially the vector layer thing, raster layers wouldn't be too bad.

if you can give me a specific example that causes layer (more likely workspace) loading to crash it would help me emencely. it's one of those things thats hard to test.

and I do plan on it saveing and loading evaluated images with the code that made them, also plan on haveing some way to set the random seed (so the random function will behave the same way everytime you evaluate it, but you can still manualy step through untill you find a seed you like,)

[Bobboau]

I updated the zip again, mostly work on getting output saved with code, some changes were made in how layers and work spaces are stored, most importantly layers and there image files (img) are in a subfolder of the folder the workspace file is in, there might be some problems with it porting the old style over to the new style, so as always its a good idea to backup, work flow should be greatly improved with this version as you no longer will need to reevaluate any layers after opening a workspace, everything should be exactly as you left it, (well exept the last layer will be selected, but that just takes a single double click to change). I also have the basic code for loading other file formats (other than the simple and prox only img file) but it isn't used by anything yet.

[Water]

a lot, especially the vector layer thing, raster layers wouldn't be too bad.

if you can give me a specific example that causes layer (more likely workspace) loading to crash it would help me emencely. it's one of those things thats hard to test.

Downloaded the latest one, haven't had a chance yet to test it. Should be able to test and write up tomorrow

[Bobboau]

I think I fixed the overwiteing when loading a layer bug, but it's not in that build it'll be in the next one I post though.

[Bobboau]

I updated the zip, it now has image layer support, so you can now use any image as a source, which means this is now useful to texture artists, especially if you want to make cool animations.

[Bobboau]

http://freespace.volitionwatch.com/blackwater/minertile2a-glow.zip

the ravana needs to be remaped for this to look right, but you get the idea

[Vasudan Admiral]

Is there a way to use a black&white image/layer as a kind of opacity mask? I've tried a couple of techniques, but can't seem to work one correctly.

[Bobboau]

if the B&W layer is BW
and the thing you are trying to mask is SRC

try
BW.l(x,y)*SRC.r(x,y)
for the red channel obviusly
depending on what is in the BW layer you might want to clamp it before multiplying

note there is a texture path in the options menu.

[Bobboau]

or do you mean you want an alpha chanel in the output file?

[Vasudan Admiral]

No - not talking about exporting alpha channels, but that could certainly be a cool future development.

Ok, now I'm beginning to suspect something might be wrong with prox. If I have image-from-file layers, I can't seem to grab any data from them in other layers.
Eg: in the layer that is supposed to mix them, I have:

@R=(EngPipe2Tex.r(x,y))*(EngPipe2Mask.r(x,y))
@G=(EngPipe2Tex.g(x,y))*(EngPipe2Mask.g(x,y))
@B=(EngPipe2Tex.b(x,y))*(EngPipe2Mask.b(x,y))

with both those layers being images from files (evaluated already), but all I get is solid black.

Then if I just have:

@R=(EngPipe2Tex.r(x,y))
@G=(EngPipe2Tex.g(x,y))
@B=(EngPipe2Tex.b(x,y))

I get solid white.
Since neither image is solid white or black, it has to be something else. :\

[Bobboau]

you now I thought I noticed some flakyness with layer lookup, try changeing the names of the layers and make sure you hit enter when your done, I guess I've got another bug to work on.

can you send the workspace? if the layers are displaying the images they should definately be getting looked up properly.

[Vasudan Admiral]

Errrm, well that was odd. I created two new test layers - one with an image and one just referencing it as:

@R=TestLayer.r(x,y)

That worked fine, and I also found I could reference this test image-layer from the older layers too. Enter didn't help with letting the older layers reference the the older image-layers though.

[Bobboau]

ah... figured out what the big problem was, there is currently a few minor... issues with layer names, they can't have numbers in them and they can't have "E" "PI" or "PHI" in them and there is no code to catch this (yet), the second part I'm going to try and fix so they can have those strings, but, the number thing is likely going to stay. guess now is a good time to fix that.

TestLayer doesn't have a number or the forbiden strings
also layers can't be named ang, r, t, x, y, X, or Y, obviusly.

[Vasudan Admiral]

Can't have numbers? Considering that the layers are saved as separate files with the layer name as the filename, not being able to use numbers to easily differentiate them is going to get tricky. :\

It would help if when saving a workspace, it saved the layers as <WorkspaceName><LayerName> or something like that. Might that be possible? (I guess separate subfolders for each workspace is another way to go about it though)

Anyways, thanks - I'll see if I can get this effect working now.

Edit: Nevermind about the workspace/layer name thing - I see you're way ahead of me there. Awesome.

[Bobboau]

I updated the ZIP, you should now be able to use E and PI ect... as PART of a name, no numbers isn't going to allowable though, you can just use layerA, layerB, ect.. to keep them seperate if you want. it's too hard to tell if the number is part of the variable name or a seperate constant, it's especaly problematic if I intent to support implicit multiplication (2x==2*x) wich I sort of would like to

[Vasudan Admiral]

Another quick question - what are you using to compile the EFF style frames into the AVI animations you've been displaying?