CAP 21 CAP 21 - 3D Model Render Submissions

[DougJustDoug]

This thread is open to WIPs, comments, and final submissions for the animated 3D model of CAP 21. At the end of this thread, the animated models will be voted on by the CAP community, just like other art and sprite stages of CAP.

However, building a fully-rendered 3D model has many more incremental WIP steps than the other art stages of CAP, and each aspect of the 3D model has specific guidelines and rules in order to yield a legal final submission. Unlike the other stages in the CAP process, collaboration on 3D model submissions is permissible and encouraged.


Base Model & Mega Evolution Model

Because CAP 21 is a Mega Evolution project, this Modeling stage will be different than a normal CAP project. All modelers/modeling teams will submit TWO complete models, one Base model and one Mega Evolution model, to make a legal final submission. The Base and Mega Evolution Models by each modeler/modeling team will be voted TOGETHER as a single submission. We will not choose a Base Model independent from the Mega Evolution Model.


Modeling Rules

Wireframe Rules

• Wireframes will be created in 3D Modeling software such as Blender (open source), Maya, 3ds Max or Zbrush. The software must be able to export to .obj (or another widely used and universally importable format such as .dae and .fbx).
• Wireframes should be inspired by the winning design from the Art Poll and the Sprite Poll (if completed). It does not need to be an exact rendition of every detail of the design; "artistic license" is granted. However, drastic deviation from the selected art and sprite designs is discouraged. The artist of the winning CAP design may provide a model sheet as reference, consisting of at least front, side, and back views of the design.
• Wireframes should have polygon counts similar to that of Pokemon models in Pokemon X and Y, which range around 2,000 to 10,000 depending on complexity. Wireframes are encouraged to have a polygon count under 20,000.
• Wireframes should be created in a neutral pose. Rigging and animation will need to be done at a later stage, so keep this in mind.

Texturing Rules

• Texture files must be saved as .gif or .png format.
• The texture resolution must be in either dimension be from 16x, 32x, 64x, 128x, 256x, 512x, 1024x, or 2048x. Both dimensions, height and width, do not have to be the same. Most Pokemon textures usually do not go over 512x in either direction, so larger dimensions should be used only if absolutely necessary.
• Shiny textures should be based off of the winning shiny sprite color scheme, if completed.
• If used, facial expression textures (i.e. eyes and/or mouth) should use a dedicated UV map and be separate from the main texture file. The map should be split into equally sized tiles, each containing a different expression.
• Other maps such as Normal, Occlusion, Specular and Reflectivity are optional, and should be of lesser priority to the diffuse (color) map, even if they are necessary.

Animation Rules

• An animation must be saved as .dae or .fbx format.
• The rig for the model should only contain bones and vertex weights (binds).
• The animation should depict the Pokemon in an idle, battle-ready pose (i.e. not sitting down or lying down unless the art design calls for it).
• The animation must be able to loop smoothly in no more than 100 frames. Physics-based animation is discouraged.
• The animation should run at 30 FPS. However, the animation may run at a lower frame rate if the final render would exceed file size limits.

Rendering Rules

• Final renders must be displayed in the gif file format, with a file size ideally lying in the 100 kB range and not exceeding 200 kB.
• Each render should be on a transparent background.
• Each render should have an outline of approximately one pixel in thickness and of a darker (but not pitch-black) version of the colors in the Pokemon's texture.
• Each render should use a cell shader and lighting set up to approximate that of Pokemon XY/ORAS.
• All gifs in a set of renders must have the same number of frames. For each pair of renders (front and back), each render should have the same resolution for both non-shiny and shiny variants.

When posting WIPs, please post screenshots or renders of your model in a compressed image format such as jpg and host them on external sites such as Imgur or Puush. Please keep these images under 640 pixels in either dimension and under 200 kB for easy forum viewing.


Final Submission Post

All modelers are responsible for making a final submission post conforming to the above rules and the following:

• The post must have "Final Submission" (in bold) as the first line, followed by four (or eight if there are gender differences) rendered GIFs: front and back of both the non-shiny and shiny variant of the chosen CAP design.
• The post must also link to all the files that were used to create the final render (as well as a model file that can be opened universally such as .obj or .dae, and the animation can be exported in .dae or .fbx format), hosted externally on a reliable file-hosting site such as Dropbox.
• The post must contain the total polygon (triangles) count, vertices count, number of frames, and the resolution of each texture map being used.
• The post must credit the people who contributed to the submission.

All final submissions may be reviewed by CAP moderators, who will consult with established 3D modelers in Smogon to ensure the model's viability.


Advice for Modelers

• Style (General) - Using PS! gifs as reference can be helpful for animation, color, detail rendering, etc.
• Style (Wireframe) - Simplify details (especially eyes) whenever possible. Eyes and small details can be added as a part of the texture map.
• Style (Texture) - Flat colors are usually all that are needed in this stage.
• Style (Texture) - There should be an outline to separate areas where different color areas meet, like in this example image.
• Style (Render) - The in-game lighting uses a directional (sun) light pointing at the front of the pokemon, with a 30-60 degree offset. This also applies to the back view, making the model appear darker.
• Best practices (Texture) - Make sure the UVs are reasonable for the texture artist to work with, such as making sure they are orderly, relatively free from distortion, and with seams placed in strategic areas away from focal points of the model.
• Best practices (Texture) - Keep the number of different materials as low as possible. Different programs may not interpret materials the same way, so a large number of materials can cause headaches when collaborating with other artists. However, it is advised to use multiple textures for textures that change (e.g., eyes or mouth).
• Style (Render) - Cameras should be looking straight ahead at the model, not looking too much above or below the horizon level. The front camera should be looking mostly at the front of the model and also looking slightly at the left side of it, while the back camera should usually be looking at the model's right side more than the back side. The field of view should be around 100-120 focal length (about 15-20 degrees of FOV).
• Style (Render) - The camera for both the front and back sprite should have identical settings (FOV and render resolution) for consistency.
• Style (Render) - It is preferred to do renders in separate passes then composite them together, instead of relying on the 3D program's built in settings. Compositing programs include After Effects, Nuke, Fusion, or Natron (open source). This optional step allows for extra detail such as multicolored outlines, special effects (e.g. fire or smoke) and other post processing (noise removal etc).
• Best practices (Rigging) - There is no limit on how complex the model's rig can be, but it should be feasible for an animator to pick it up and work with it.
• Best practices (Communication) - Update the thread frequently with what you're working on by providing screenshots, renders, or files. This allows for efficiency and avoids the possibility of repeated work.
• Best practices (QC) - It should be in modelers' best interest to post feedback for others whenever possible. This is a good way of showing appreciation for their hard work and ensures the product is of good quality.

Art Designs:

Base Forme

Mega Evolution

---

CAP21 so far:

Typing: Rock/Poison
Abilities:
Base - Regenerator / Mold Breaker
Mega - Magic Guard
Stats:
106/105/65/75/85/104 > 106/135/75/85/125/114

Leadership Team:

jas61292 - Topic Leader
imanalt - Typing Leader
DetroitLolcat - Abilities Leader
sparktrain - Stats Leader
HeaLnDeaL - Movepool Leader

Concept:

Name: Typing Underdog

General Description: A Pokémon which utilizes an undervalued typing to its full potential, by playing towards both its strengths and weaknesses.

 

[Quanyails]

This post is approved by DougJustDoug.

CAP 21's model sheet:

Base forme:

 

[DJTHED]

Okay, I've already go a wireframe to show off, so here it is.

Crucibelle-Base Wireframe WIP
Screenshots:

Hide

The total Polygon count is at 4,380 (2,454 verts). The model is pretty simple. Most of the complex detail will be done in the texturing phase. Yes, including the hair. The hair texture will wrap around this:

Hide

Yes, I know, it makes it look like a beekeeper at the moment, but the texture will have transparency so that you will still be able to see through any untextured parts of the hair. In my opinion, it's better to model it like this and have a texture wrap around it than to model the hair completely for two reasons. For one, it prevents any unnecessary increasing in the polygon counts. I'd rather not model something with too much geometric detail (at least for this project when we're trying to match the style of the XY models) unless it is absolutely necessary. And two, it will allow the texture artists to draw the hair in any way they see fit.

And just so that I can keep everyone's sanity, here are some screenshots without this showing so that you can see the head clearly:

Hide

Link to .obj and .blend here. I plan to improve this much more and clean it up. Especially the hands, which I'll need to make wider than the arm just a bit and better prepare it for rigging.

 

[QxC4eva]

:D That easily looks ready to go! Uh, maybe scale up the hands a bit? I would also trim down the belly fat (the side is fine, just bulge at the front) so it's easier to get those curves on the reference art when you pose it.

But yeah, it's a very minimalist model so I won't nitpick. The mesh looks clean and the polycount kept low, so good job.
Are you going to do the UVs? I'd like to help out on that. =)

 

[DougJustDoug]

The hair mesh should bend a little bit towards the head at the bottom, making the mesh shape have slightly rounded sides, not perfectly vertical sides. This is consistent with the model sheet (although the art is fairly straight) and might make the hair look a little more natural. My knowledge of fluid dynamics is limited, but I think there would be some surface tension making the hair goo cling a little bit to the head as it hangs/drips down.

 

[DJTHED]

Okay, here's an update:

Crucibelle-Base Wireframe WIP
Screenshots:

Hide

Changelog:

• Worked on the hands a bit more, and enlarged them a tad.
• Less belly fat at front.
• Hair curves inward at the tip (I believe this is what you mean DougJustDoug)

Download

Are you going to do the UVs? I'd like to help out on that. =)

Yeah, you can do them. You're clearly better at it than I am anyway. :p

 

[QxC4eva]

Okay I'm happy with that! I think the arms can be a bit longer but you can decide. :P Maybe add a bit more padding on the hands? Once that's done I'll start the UVs!

DougJustDoug I like the surface tension idea :D Could be something to keep in mind for the animation stage.

 

[DJTHED]

Crucibelle-Base Wireframe Final
Note: This is NOT a final submission post.
Screenshots:

Hide

Changelog:

• Smoother curve from the arm transition into the hand at the top.
• Made the arm longer to approximately the same length as the Art Design.

So that's that for this wireframe. If anyone else has any more QC suggestions, there's still a window for QC right after QxC4eva is done with the UVs. I'll be able to make small minor revisions then, but if there isn't any by then, then this should move on right to the next step.

EDIT:
Lol, I completely forgot to insert a download link for some reason. :I

Download

 

[QxC4eva]

UVs are done!

• UV map: http://i.imgur.com/joXRTS0.gif
• Seams: http://i.imgur.com/rhS0o0a.gif
• Checkers' test: http://i.imgur.com/cUDmJHB.png

Here's the obj.

Alright I'm gonna start blocking out the texture map and do some test renders. Fyi, I'm coding up a new plug-in just to handle the outlines for the hair. If someone wants to do texturing, please hold cause I might need to tweak a few UVs to get it working. Hopefully we get to see something soon. :D

Credits so far (for this revision)

Model sheet: Yilx
Wireframe: DJTHED

 

[QxC4eva]

Crucibelle (non-mega) Texture / Render WIP

I blocked out a rough draft of the texture map so we can get a preview on how the render might look. There's no actual texture yet it's all flat fills right now. XD The hair outlines aren't rendering properly as expected. The plugin should be finished in a day or two, and hopefully by then hair won't be a problem anymore. :P

And here's a render in default settings: http://i.imgur.com/9ZIEM8Y.png

 

[KrazyCake]

Umm hi guys I just noticed something if I am not mistaken the texture for crusibelle has two eyes but in the artwork it has one this is quite a big feature because in the mega it has an eye patch.

 

[DJTHED]

Umm hi guys I just noticed something if I am not mistaken the texture for crusibelle has two eyes but in the artwork it has one this is quite a big feature because in the mega it has an eye patch.

I think it should be assumed that its other eye is being hidden behind the hair. Yilx can you confirm?

 

[KrazyCake]

I think it should be assumed that its other eye is being hidden behind the hair. Yilx can you confirm?

Hmmmn yeah Yilx can you please confirm that. The reason why I am so sure that it has one eye is because it is shaded in the place were the eye should be. Implying that the eye is not there and also da mega is rockin its eye patch

 

[Yilx]

Yeah, it has another eye hidden behind the 'hair'.

 

[QxC4eva]

Crucibelle Texture / Render WIP

I got the hair working! :D Still have to fix up some bugs then it should be fine I think. Basically how it works is for each pixel making up the hair, it samples the four adjacent neighbors, and works out the vector between their U and V sample spaces. If the vector's magnitude is above a certain threshold (meaning they're far apart in the UV map, suggesting a front part of the hair is blocking out the part behind) then it's an outline and the pixel is darkened. It also does extra samples with +1 and -1 offset, and uses the lowest distance just to avoid drawing an outline on a texture seam. For the edge boundary of the hair, the z-depth between neighboring pixels is compared and an outline is only drawn if the hair pixel is in front.

I still need to tidy up the texture files so I'll post them tomorrow. For now if anyone wants to do texturing feel free to use this as a template:

http://i.imgur.com/dGmMfdc.png

 

[Mega-Pokebattlerz]

Hey, nice work so far! I think the model looks decent and the texture is alright, so maybe I could work on the animation aspect. The hard part will be getting the slime to drip, so I'll use Goodra for inspiration.

 

[DJTHED]

I just feel like I need to point out something with the texture.

In the texture, the indention in the lid/helmet is at the center at the very front of it, while in the art, it is clearly right next to the left spike. I'd also assume that there is a second one that's mirrored on the other side of it, but just out of view due to the perspective. I know the model sheet doesn't show it like that, but this is what the art indicates, so... Should we consider changing that?

It would go somewhere around the red circles are instead of the center (perhaps a little smaller too).

 

[KrazyCake]

I just feel like I need to point out something with the texture.

In the texture, the indention in the lid/helmet is at the center at the very front of it, while in the art, it is clearly right next to the left spike. I'd also assume that there is a second one that's mirrored on the other side of it, but just out of view due to the perspective. I know the model sheet doesn't show it like that, but this is what the art indicates, so... Should we consider changing that?

It would go somewhere around the red circles instead of the center (perhaps a little smaller too).

Yeah that does sound about right but again I think Yilx should tell us

 

[Yilx]

Yeah that does sound about right but again I think Yilx should tell us

That works too, probably a drawing error on my part. My apologies

 

[QxC4eva]

Oh I was going off what you said before but okay. Fixed!

Crucibelle Texture Final

Here's the updated files. I added new UV sets for the eyes and mouth so the model file got an update too. It's ready for rigging and if there's no takers I'll start it tomorrow. :)

Other credits

Model sheet: Yilx
Wireframe: DJTHED
UV Mapping: QxC4eva

 

[DJTHED]

I think I'll tackle the rigging and binds and hopefully have them finished no later than Monday. Though the FBX you provided doesn't seem have the proper UVs for the mouth and eyes, at least that is the case when I import it into Blender.

Also, don't forget to outline these parts in your final Maya render!! (Ignore the crappy hair outline. Blender is dumb.)

I hope those two first pics you posted aren't the final camera positions for the front and back sprite. The camera positions need to be a bit more level with around the mid-body, and not way above its head, looking down on it. The camera should be looking straight ahead, not looking too much above or below the horizon level. The field of view appears to be okay though (which it should be around 100-120 focal length, that's about 15-20 degrees of FOV). Render resolution also appears to be okay, though it is just a tad large imo.

This also got me thinking of something that should be added to the render rules in the OP. For example, the camera for both the front and back sprite should have identical settings (FOV and render resolution) for consistency. The front camera should be looking mostly at the front of the model and also looking slightly at the left side of it, while the back render should usually be looking at the model's right side more than the back side, as most of the PS XY back sprites are (I didn't quite do this too well with some of my previous renders, which I would like to fix some time in the future). This should be added to the OP as well as the FOV settings and the other points I made about how the camera should be looking at the model that I stated in the previous paragraph.

EDIT:
Oh, also, I think that in the texture, it should have more of a subtle "earthy brown" instead of a near colorless black and grey, as Yilx had done with his art when it was still a work in process.

 

[QxC4eva]

I did forget, thanks for reminding me!!

Here's the updated texture file.

Changelog:

• Added missing outlines
• Made the grey earthy brown

Hmm I looked inside the fbx and all the uv data is there. I also tried importing the file back to Maya 2016 and Max 2014 and both works ok so it might just be a Blender thing, maybe check on the settings? I tried exporting to dae as well but the eye and mouth uv don't seem to be there, Maya messed up that one so I can only use fbx. (so many dumb programs -.-)

They're not final cameras, I think we should only decide the cam angle after the posing or animation. Also yeah I notice the OP has nothing about cameras :P You have some really good points there, maybe they should go as tips instead of rules? I think artistic freedom is nice... if something looks off you can always beat them up in QC. :D

 

[Yveltal]

Hey guys, quick poke:

How were the cameras set up for Naviathan? Whoever has the files with the final exports for its animations, let's take that and create a scene with it to load a completed model in. We can make that the standard to render from.

 

[DJTHED]

Crucibelle-Base Rigging WIP
Consider it done if no one finds any problems with it.

Hide

And here's a test animation showing the rigging off a bit. (Btw, I did manage to find the UV mapping data for the eyes and mouth, but I had no idea how to get it to work in Blender because the UV data for them is also already being used by the main UV map.)

Beautiful

One thing I should note. I had to shorten the arms, basically reverting them back to how they were originally because they were way too long and made it impossible to pose it the way it looks in the art. But the wireframe is otherwise the same.

The download link for the zip containing a .blend and .fbx file is here.

 

[QxC4eva]

Nice entry animation :D

Make the shoulders shrug a bit more I think. The rig looks good, I know you're going to animate it so feel free to start! Oh and can you also send me the idle pose (the one you'll use for animation) so I can work on the render? It doesn't need to be final I just need something better than a T pose :)


edit- ahahah

<-- he looks quite funny like this :P