Be careful. Going gray can cause issues. Some sprites u have to split into 2 layers, one for grays that get blended and ones that stay true. I almost ruined the LPC spiders this way

Another note...speaking of knowing insights of the sprites, the handedness problem bothers me sometimes. Two frames are left handed and 2 are right handed. And the front facing sprite is one of the left handed ones, which is weird to me. Something I have done in my engine is to flip the X on every front facing sprite, so 3 directions are right handed and only one is left. 

Why are you detecting all the colors in the sheet? I just simply use a color blend (with option for alpha which lets me add ghost items) . Sounds like you're trying to recolor manually? 

For every draw category (hat, pants, torso, etc) have a corresponding color stored as rgba ints or floats. Default them to white (no color blend) and apply the blend to an entire blit

My code isn't stable, it breaks often.

I mostly implement features on a very basic level, but I would need to do some polishing later.

I've implemented color palette switching.

So I just override the name and switch to new color palette. This means, colors are limited to available palettes. Palettes should be chosen or at least fixed manually, which I have no experience with. It looks better if there aren't so many different colors used, also automatically recoloring with gimp changes colors of eyes and the outline.

Since I do simple color replacement, I have to specify the used palette in tsx files.

I modified some images, since I'm not satisfied with the current quality. Light male used 82 different colors, which I reduced to 12. I also shrink it all down to 8-bit png.

I think about splitting base into skin color and body type selection, since all clothes depend on the body type.

Best default color is light skin tone and white, I guess. Darker palettes merge some colors, which would be bad for replacement.

Flipping down fucks up the reflections. Handedness is inconsistent, because left and right are nearly just flipped (reflection was the only difference I saw).

Good point about the flip. Now that I see the light issues I can't unseen them. Plus it violates a basic rule of LPC (where to assume light comes from on sprites)

 Palettes are definitely the best way to go about recolors, but I find for a lot of basic things, color blending a desaturated spritesheet to work great. I use it for shirts pants shoes gloves cape capetie. For Cape clip I have two sheets, one for the base and one for the colored clasp. Both can be color blended different colors.

it doesn't work as simply for shiny metal armors. The whites have to be saved in their own spritesheet to avoid color blending the reflection. This is still far simpler than palettes and mapping

Still, some pieces just can't be color blended this way without destroying quality. But when it does work I find it simplifies a lot and saves a ton of work

I'm getting ready to release an armor pack and wanted some feedback before I post it

Meant to be: iron, steel, gold, brass, bronze, copper

Also fills out 2 styles of helmet.

I was also thinking about the ability to recolor the sheets in the generator. If we grayscale all sheets, and make some color picker in the generator, this might be doable. Maybe like how they do it in the online Lottiefiles editor, it should be doable and will deliver a lot of value.

Being able to recolour assets would definitely be useful, because it's such a tedious thing to do by hand. That said, it's tedious precicely because it's not as easy as convoluting a greyscale image with a colour.

A lot of (original) LPC artwork is based on 5-shade ramps and can be recoloured by swapping our colours for the corresponding colour in one of the other ramps from the LPC palette. To keep a consistent style, it would be best to pick colours already defined in the LPC palette rather than picking a new set from scratch, if possible (I'm not particularly fond of the LPC palette: it has a lot of green and brown shades that are sortof but not really interchangeable, and it's lacking in bright reds and yellows, so you need to add those if you want them).

Now, not all artwork that claims it's "LPC compatible" follows the LPC palette. That's ok, there can be good reasons for using colours not in the palette, but sometimes it doesn't use any of the "LPC palette" colours, and it looks off when used with other LPC artwork. At least it does to me.

Copper should look more brownish, I think.

Brass looks flat.

Those are likely issues who where there before:

Steel alone uses 34 different colors (with background). That shouldn't be. Original plate armour "only" uses 13 colors, which already pretty much. For comparison: base characters use 11 colors and this includes eye color.

So you use basically a multitude of palettes, the horned helmet uses completely different colors from the rest.

The outline border within metal isn't consistent (horned vs. rest). I would maybe use black for iron and bronze and back to #281820 for the rest.

And I don't understand why there are two versions of the horned helmet, one shiny and one dull.

---

A few assets do not look anything like LPC despite tagging themselves so.

My own stuff doesn't look better. For triumph I just colorized it with gimp and picked the colors from the original captain triumph imag. I tried to use the colors from the base asset for my police car, but it looks weird.

I agree with the brass. Too dull, right? I'll see if I can fix it. I was too afraid of it looking too close to the gold, and too close to the Roman bronze (which looks more like brass than bronze to me). As far as gold shoulders, I already remade the shoulders without the spikes. As far as the ornate helmets, I did not add a death scene. (Yet) I'm not a huge fan of the death cycle, but I will give it a try.

I think my metallurgy is too based on modern plumbing and not enough on ancient and medieval metallurgy.

As far as the steel colors go, I'm not sure I understand. Are you saying I added more colors to the steel? I didn't think I modified it, the steel should just be the default (except the more ornate helmet) plate. It sounds liek you're saying technically it would be better to have more universal paletting, but that its not always realistic? I'm a little out of my element here, is it something I should really try to fix before releasing any of the armors?

@castelonia: as I sai before, somebody fucked up the base sheets. I already noticed this when I with my full body suit, and mentioned it here before.

I just looked into it. It's because of the new animations.

With shooting added it goes from 11 to 20.

With piercing added it goes from 11 to 75.

It is still basically just 11 colors, since there are many colors are just looking slightly different.

You can use those from my repo, normal and muscelar male as well as normal and pregnant female use a palette of ten 11 to 12 colors. It's only indexed 8-bit in my repo, with canvas it gets converted back to full color. (only light skin in repo)

The eleventh color is not used yet, but is intended for the teeth of orcs. Grayish orc has 159 colors, light green orc has 244 different colors.

With < 20 colors it's doable, but I would still recolor automatically then just fix what looks weird. Or pick colors from the base palettes or so. Having more than 20 colors is rather a sign for having too many colors, I think.

It would make sense to bring all characters light palette and all clothes maybe yellow. At least colors which aren't too dark or to bright to see the details.

With that base you can "just" switch the color palette. Before doing it with javascript I just did it with mtPaint, which can load gimp color palettes. You will only see a change with indexed images, but when you convert to indexed it allows to "use current palette"

@bigbeargames:

No, I think the images you started working with aren't good. It's likely the same for each color, it has so many colors mtPaint didn't even give me the option for an exact conversion. That's why I focused on iron.

I think a problem is that the sprites you use have different colors to begin with. (horned helmet sticks out)

Other tools should be possible to limit the used color palette as well, mtPaint is just what I'm working with, but it's a dead project.

The original plate armor is good, but I don't know where the horned helmet is comming from.

I'm starting started to bring all iron and metal stuff I can find onto one palette (7 colors + background), but I think I'm using too much detail on the reflection. I just let it do mtPaint automatically, which might not be the best idea.

I'm using the iron palette of roman plate.

The horned helmet is the original gold helmet. I sampled colors from the original plate to color it in to make the steel one, but I may have missed some similar looking colors. I noticed the gold plate set to not just be a recolor of the regular. My goals here aren't so much artistic as to flesh out the LPC sets to fit the needs of a full paperdoll in RPG. It is my hope that some of my hacks will function well enough to be later fixed by people who actually know what they're doing. For now, it satisfies my untrained eye and I'm hoping it will satisfy many other untrained eyes.

I'm halfway done with my hair masks. I decided to do a mask for each hair/helmet combo (many can be repeated, butI still ended up with 9 masks per hair style. I selected 20 hairstyles to work with (I left out some hair styles that are more complex female hair) For the two closed visor helmets, no hair is drawn at all. If anyone wants these at any point let me know.

Here is my commit, I'm now mostly satisfied. Spiked helm needs reflectiones improved. Gold is so bright that it loses detail there, I also mapped all light colors to one for gold palette. I included two examples for gold palette (from original gold mask)

I just mapped colors and used the 12 color palette of wulax. This means there is no color loss for the knight amor. I think it's too many colors, but this palette is so carefully crafted.

You could also include legion armor, tiara, buckles and belt (at some point later). Those even include gold and bronze already.

But in fact, a player playing an LPC based game will not notice 95% of the issues we are spotting.

I look at the color count ... it's not really something you can see at all. I just like the limitations this stuff comes with.

From the begining there were assets which appeared to not fit. It was a jam, they only had limited amount of time. I can't do it any better, but the more I look at it with 1200x zoom, the worse it gets.

Skorpio's pack has some cool stuff, but other stuff that just looks bad. And yes, better than nothing. Otherwise I don't think I would've ever uploaded anything. I've other stuff lying on my disk which could theoretically be used for static NPCs in games. But it's not finished enough or looks too bad from one direction.

Also I often had trouble to get the attributions together, which hopefully gets easier some time in the future.

Back to  topic:

Why lack Dark and Dark 2 a belly button?

And why has dark green reptilian light green wing in some directions and vice versa?

No idea what's up with the belly buttons, but the wings were mislabeled in the original submission.

Earlier in the thread we discussed dynamically placing female head-based sprites. I have come up with a list of the head-to-head transformations from Male to Female:

Please note frames start with 0 for first frame. Direction order: up left down right

Up: Set female's Y as Male's - y

Down: Set female's Y as Male's + y

Left: Set females X as Male's - x

Right: Set females X as Male's + x

cast: good to go

spear,dir0,frame0: down 1

spear,dir0,frame1: left 1

walk,dir0, EVERYFRAME: down 1

walk,dir1,frame2: down 1

walk,dir1,frame6: down 1

walk,dir3,frame2: down 1

walk,dir3,frame6: down 1

slash,dir0,frame1: up 1, left 1

slash,dir0,frame2: up 1, left 1

slash,dir0,frame3: up 1

slash,dir0,frame4: up 1

slash,dir0,frame5: up 1

slash,dir1,frame1: right 1

slash,dir1,frame2: down 1 right 1

slash,dir1,frame3: right 1

slash,dir1,frame4: right 2

slash,dir1,frame5: right 2

slash,dir2,frame1: down 2

slash,dir2,frame2: up 1

slash,dir2,frame3: left 1 up 1

slash,dir2,frame4: left 1

slash,dir2,frame5: left 1

slash,dir3,frame1: left 1

slash,dir3,frame2: down 1

slash,dir3,frame3: left 1

slash,dir3,frame4: left 2

slash,dir3,frame5: left 2

shoot,dir0,frame0: right 1

shoot,dir0,frame1: up 1

shoot,dir1,frame0: left 1

shoot,dir1,frame1: left 1

shoot,dir1,frame2: left 1

shoot,dir1,frame3: down 2 left 1

shoot,dir1,frame4 - frame12: left 1

shoot,dir2,frame0:right1

shoot,dir3,frame3: down 2

death,frame0: down 1

death,frame1: left 1

death,frame2: left 1

death,frame3: left 1

death,frame:5: left 1

I have tested this and it works great for all my hats, helms, hair, facial hair, glasses, nose, ears, eyes, and lizard gear.

It was enough that I plan to remove the female assets in those categories and simply apply the transformations. Can anyone think of a scenario in the future that this wouldn't work for? With the exception of a few female-only tiaras and such, I believe all male and female are the same hats, just placed differently.

It does not work with all headwear.

My mask, which is basically a recolored head, won't work.

Ears, nose (side view), eyes and chin differ, everything that does not cling tightly to those areas should work.

I just noticed, that elven ears sometimes contain an eye color pixel. That shouldn't be.

In the past I used the generator mostly for finding assets again, I already know. For me it was much quicker than going through the collections. Also sheets got merged with all extenionsions.

It's likely that there are quite some additions, that were never uploaded to oga.

Giving more detailed attributions and relationships of derivative work is very important to me:

• It must be giving credit to every work a sprite sheet is based on
• Currently there is often no way to propagate changes to derivatives

Right now with generators and spritesheets it's often so, that you have to go carefully through the authors list or even list all possible authors.  You can't find the originals, can't comment on the uploads, it's hard to get in contact with the authors.

It's hard to confirm the attributions or reconstruct changes. Some attributions are pretty vague. Since your fork is based on the generator, you've cut off the history of Universal-LPC-spritesheet which likely makes it even harder  to find out who did what.

It would also be nice to include unfinished, very limited and not properly fitting sheets. Which should be disabled by default and only enabled with expert settings. Same goes for showing sheets that don't match the body shape.

I'm not sure if I want to make the 'none' body also an expert setting, but I think it's neat when you just want to combine some clothings without creating a full character sheet.

Maybe even adding a tab with a list of complete characters (sara, laurelia etc.) could be neat. They would indeed be only character links. But that could also be done via readme.md or so.

I did implement masks without testing or using it yet, but it should work. It just sets the alpha channel of the image to the red channel of the mask right now.

Next thing would be to do animation and previews, without repeating the errors of the original. Children look pretty broken right now.

After that relations should follow. I started using it in the tsx of roman amor, but it is not displayed yet.

does anyone have a texture atalas for these sheets that can be used with the  libgdx texture atals class? if I not I can make one  but i'd rather not reinvent the wheel.  Also did any one ever make a backpack to go with these? not a quiver but  a backpack?

If you are going to make an atlas, it will work a lot better if you split the massive 6 cycle sheets into smaller sheets so they puzzle better together in the texture packer for libgdx (https://github.com/crashinvaders/gdx-texture-packer-gui). I tried making atlas for the sheets and found no performance improvement in loadtime, but a tiny speed impovement during render

I was going to make a backpack but never did but now that I know there is demand for it I will try to make some.

I wanted to leave the massive 6 cycle sheets so I could easily change armor  and weapons at run time  and randomly generate enemies / soilders to recrurte/ NPCS  that all look different. Also would there any easy way to take these sprite sheets  and add throwing objects anamations   picking up up objects anamations  ,  and eating anamations. Or would that be hard?

You can do both. In my game I also have randomly generated enemies that can have many different gear, and I split the mega-sheets up too. With photoshop batch processing or many other graphics apps, you can split it pretty quickly.

Any animations you add will either require being updated for a LOT of gear, or you will have to hack it using different pieces of different cycles. I am in the process of adding horse riding with similar hacking. For one frame I copy the top half of one sprite and the bottom half of a different one to produce a new effect, and the same hack can be applied to every piece of equipment too.

I wrote small java desktop  app for  splitting up images. It does it   very quickely now I'm  writing one for naming them quickly too .  Including bitmasking for building / terrian tiles.

Awesome I'd love to see the code sometime.

Very cool to see some action on this again! Two suggestions:

• ElizaWy was working on a completely separate character creator here: https://opengameart.org/comment/64024. She did a lot of work a few months ago on auditing the character sprites; she was tracking progress on that work here: https://docs.google.com/spreadsheets/d/1kME4PrOuKnOx9KSaVw_c2qwAbdLljCrC... . Unfortunately her health seems to have worsened and she hasn't posted in a while. I don't think a version of the character creator was ever uploaded.  The most recent version she uploaded of the sprites seems to be here: https://github.com/OpenGameArt/LiberatedPixelCup/tree/master/sprite/char...
• Please upload any modifications/additions to OpenGameArt in addition to making changes to your forks of the editor or universal spritesheet repositories! We've tried a few times to come up with a more "definitive" or "curated" version of the LPC assets (see e.g. here: https://opengameart.org/forumtopic/suggestion-github-organization-lpc-te... ) but what we've found is that OGA is the most reliable and discoverable resource in the long-term. Also post early and often---even if something isn't quite done! Otherwise it's too easy for things to get lost/forgotten about (I have this problem all the time where I'll lose interest in pixel art for a few months, and my projects lie stagnant on my hard drive...)

FWIW, the most comprehensive collection of LPC assets is here: https://opengameart.org/content/lpc-collection . elmerenges, AntumDeluge and I try to add stuff there as it's posted, but let me know if there's something you think should be added. 

code and jar for image split program can be found here https://github.com/macmanmatty/png-split/releases

Cool! Yes, there's some more information on her blog here: http://www.thelastscore.com . I think it was meant to be a standalone app as well that could generate spritesheets. Either way, more choices is not a bad thing.

Did you find anything that was missing from the collection? https://opengameart.org/content/lpc-collection If so, could you please let me know so I can add it?

I removed nearly all graphics from my fork.

I plan on seperating graphics and code again.

If done right, the spritesheet generator could also be used with other assets. Only question is whether that will work with github pages then.

The only thing hardcoded is the position of the first list.json and the sexes.

I also nearly got it running with nodjs now, thanks to node-canvas, xmldom and pango. (Palette switching is missing)

./cli.js test.png --sex=1 --body="body.light"

I'm rather "again" on it.

Curious to see how you will want to make it work with other assets as well, as the current generator has  some asset specific logic in it (like the oversize drawing and eg the hair layers for some hair types).

My generator does not support that. Dynamic tile size will be a headache. MV spritesheets also break with their tile size for "dmg air" and "KO" animation, so I guess that is a needed feature. It will definitely not be possible to mix tile sizes in a source spritesheets and it will blow up the size of the generated spritesheet.

I'm even lacking preview images and animations, this comes when I'm able to automatically reposition tiles.

I'm planning on including the generator in the spritesheet collection, not the other way round. That would mean adding the generator as a submodule and providing an adapted index.html. It looks like this could work with github pages, but haven't tried that.

Big weapons work with animations. The generated spritesheet has the frame dimensions of the biggest source sprite sheet. Smaller frames simply get centred. I dislike spriteeshets with non-uniform sizes.

Preview

Asset specific behavior will be specified in animaitons.json, list.json and the .tsx files

By now I could use assets, which have seperate files for each animation like medieval fantasy characters. I want to keep all assets unchanged and only add meta data. Every asset that needs change (colors etc), will need to be reuploaded to oga. This was very intransparent before.

I have yet to find a way for more complex filters. I think hair shadow will fall into that direction.

Right now the only filter is sex, but nose and ears also need skin color filters.

EDIT: Filters are dynamic and ears are working. Their preview has the color of the currently selected body. Custom properties for this look like "filter_body=1" (light) and "filter_sex=3" (male and female). A flexible system, but very unhandy for setting it up.

The filter is like this, I think hair could work with that:

{"name": "body","values": ["light","dark","dark elf" ]}

With filters I finally managed to seperate body shape and skin color:

The body list of this generator is huge and has much redundancy. Children don't even work properly.

With palette switching it's also much less work. You just need every body shape in light. Put them all into a list. Load this list with multiple different palettes:

"body/",
{
"file": "body/",
"filters":{
"skin": "2"
},
"palette": "skin/dark"
}

My $0.02 regarding sprite sheet size:

I can't speak for other developers, but I'd be hard pressed to identify a situation where I need all LPC animations for a single humanoid character -- I always end up cutting it up into frames and only using the ones I need.

For that reason, I'd suggest the option of only exporting the animations you want (either in a strip or as separate images).   If that's too complicated, I'd suggest keeping them in their "standard" frame size...not much benefit in having them all in larger frames.

It would also be easier to add custom or new art to a generated sheet/character, most of which is formatted in the "standard" size.

I started my own spritesheet collection

Compared to the other generator it's extremely slow and locally it does not even switch all palettes as it should. It's loading a huge amount of files without caching, since I often just reference other files.

Likely quite some lag comes from all the reordering and palette switching. I don't even have much in it yet.

I haven't fixed spritesheets with wrong colors yet, since I'm to lazy to upload them to oga. But if that would work I would have body shapes with skin tones: normal (m/w), child, teen, muscular and pregnant. Included animations for male so far: spellcast, walk, slash, thrust, shoot with bow, shoot with guns, run and hurt.

It's possible that it gets unusable over time, unless I manage to get an performance boost or do preprocessing.

It could also be cool for locally testing new assets against the existing ones.

And I will not add, what I can't attribute. I could nowhere find round ears...

Children do work in the generator, but they only have walking animation so you dont see a proper preview.

I meant, that the preview is broken for everything that does not have the animation, the preview uses. My generator on the other hand requires that the first line is down-facing.

Maybe, could it work to instead of saving the complete sheet, save it as with multiple files with the animations seperated? I dont know if it is possible, but it could be a solution.

I thougt about implementing than into the nodejs version, too. I could use multiple canvases in general. I think that would also make it easier to hide sprite sheets.

Workaround for the terminal could be to add "--animation=<file>" and specifying there which animations should be exported.

bigbeargames, just wondering if you were planning on releasing your pack of armor tweaks and recolors anytime soon. I'm working on some shields with castelonia and it would be helpful to compare to/harmonize with your palettes. 

 thanks for the wonderful job ... is it possible to insert the option to modify the sprite size? It would be nice to create bigger enemies than the hero to use as a boss

thank you ... I also include a program that you can find here https://code.google.com/archive/p/2dimagefilter/downloads 

however it still requires work on individual sprites ... can't you just add this feature? it would be wonderful and you are very good!