Need some feedback on a character

Jarjarniks

Hey all! I am a struggling 3d student, on my second year of studying 3d for game art, and I find myself in need of some technical feedback/advice.

About a month ago I started work on what would be my second game character ever, and I have now gotten to the texturing process, and am starting be unsure of what exactly I am doing. Texturing is what I am worst at, and I could really use some help.

So the problem I am having is this:

I want to showcase this character in marmoset toolbag, and am currently texturing it in DDO. My character is split up into 4 different maps:

1 for her head, hair, mouthbag, teeth, tongue, eyes and eyelashes.

2 is for her body, arms and legs.

3 is for her armor

4 is for her weapons and accessories.

All of which are 2k in size.

Now heres where I am unsure of what to do.

What are the current standards of texture sizes for a main character in a video game nowadays? Say the ingame action was 3rd person over the shoulder view, like in the Batman or Witcher games, with ingame cutscenes of character closeups, how many different maps would you make for a main character? does the 4 different ones I have sound reasonable? Is 2k resolution completely insane if there is Diffuse, Spec, Gloss and Normal for each map, (of which there are 4) for current platforms as the 360, ps3, ps4 and One consoles as well as pc?

The second question I have is this:

For her hair, I want to use transparency, but is this a bad idea to do if the hair planes are located on her head map? Will I see any issues with the head etc. becoming transparent too? Should I have the hair on a seperate map from the head?

Heres a picture of the head map so you can see:

http://www.polycount.com/forum/attachment.php?attachmentid=20077&stc=1&d=1412598196

The hair planes are at the bottom of the map.

Anyway, I hope this didn't come off as too rambly, I am just feeling a little frustrated and lost, and finally decided I should ask for help from the pros.

If any of you took time out of your day to read through this post, thanks alot!

Regards,

Nikolai.

Dr.HeatSync

You sound like you're in a crisis of not having established a target platform. What generation/system do you want to target? PS3/360/WiiU can be grouped together, and you've got the current top of the chain with Xbox One/PS4.

Decide on a platform group, and then research the rough kinds of specs you'd get; For example if you want to go for 8th gen, this slides suggest the limits that the pros behind Killzone: Shadowfall think for their characters:



GDC Slideshows are pretty good for this kind of stuff, as you can see KZ:S has 2K textures for each part. Those two high end systems have around 8 gigs of shared ram. Looking back at the 360/PS3 era where you had around 256 megabytes its a stark contrast.

Mind if your textures have less detail on them (e.g. gloss maps, maps that have more flat colours that aren't reliant on a high resolution) you can always scale them down.

Its also worth remembering that the importance of the asset dictate how much of the resource pool it should take up; a protagonist is gonna take up a fair chunk (but not overkill), whilst some grunt who gets taken out by them in a second is quite tight on its expenses.

If you really wanna go deep you can look into the different types of texture compression: mainly DXT1/DXT5, BC5 (otherwise known as 3Dc/DXN for normal maps). It is quite deep for artists but in engines like UE4 your TGAs are usually converted to such formats and have different sizes to go by.

At the end of the day though, what really matters is that the asset looks good for its purpose so really concentrate more on making an asset look good; efficiency as you do more research and try more things.

---

Wall over, for your second question this is admittedly why I prefer rendering in a game engine rather than something like marmoset because they often have annoying quirks, but I think they're worth learning.

I don't use marmoset but I do recommend that hair should be on its own map; this is because of a few things:

A) The more opacity map pixels you have, the more pixels that are having to be checked for 'how transparent am I?'. If you have parts that are always opaque, they should be on their own opaque material/shader

Some engines (hello UE4) don't necessarily do great translucency; you might find translucency sorting issues so you should have the hair on its own map; the last thing you want is the hair rendering through your head. I think marmosets a lot better for this stuff though.

C) You'd probably like to have anisotropic shading on your hair, but not on everything else. Putting them on their own shader makes that very convienient if Marmoset does Anisotropic.

D) Your head on a 2048 map gets to use less than half of that; thats pretty small if you want closeups. I think it would benefit your head greatly to allow it to use the full 2048 and have your hair on a 1024 or 512 map

Finally, the best way to learn is to do it and show us: You haven't shown a lot so I can only bore you with some technical talk. Get some renders of what you've done out, high poly, low poly, what have you. Sorry for the massive, post of boring. Good luck with your studies and I hope your model turns out great

Jarjarniks

Thank you ever so much for the info, it would see you hit the nail on the head with regards to me not having a targeted platform, gee whiz that was stupid.

Okay, I am going to take a bit of time to re unwrap the head and place the hair, her eyebrows and eyelid hair on a seperate map to increase performance, that seems like a good way to also get much more resolution for her face.

I am going to keep making the texture sizes at 2k (for the big pieces), that doesn't seem unreasonable for current gen, so I am glad I finally set that in stone. In any case they could be downsampled to 1k for lesser systems, and It would probably still look alright? From the small amount of texturing I have done it would seem that at these resolutions it is not so feasible to have a lot of small scratches and battle damage as it will end up looking blurry? I am not sure.

For more info the character herself, she is based on one of bogdanbl4's (whom I idolize along with Hazardous) character concepts, here is a link to the artwork on his deviantart page: http://bogdanbl4.deviantart.com/art/Archer-concept-396338094

I created most of the high poly details in zbrush, which can be seen here:
http://www.polycount.com/forum/attachment.php?attachmentid=20078&stc=1&d=1412606484

http://www.polycount.com/forum/attachment.php?attachmentid=20079&stc=1&d=1412606484

http://www.polycount.com/forum/attachment.php?attachmentid=20081&stc=1&d=1412606484

http://www.polycount.com/forum/attachment.php?attachmentid=20082&stc=1&d=1412606499

Its the second time I create a high poly version of a character for baking purposes, I am pretty happy with it, my main inspiration for the armor inscriptions were celtic symbols .

I'll post back with a couple of pictures of the low poly mesh soon.

Regards,

Nikolai.

*Edit* On a related note, what on earth is anisotropic shading? I have never heard of it before, but I guess its only a google search away

**Edit** Also, I have now set her eyes and caruncle on their seperate map, what sort of resolution for eyes could be seen as reasonable? I downloaded the recent Lara Croft from Tomb Raider, and I see crystal dynamics also set her eyes on a seperate map at 256x256 texture, but this seems awfully low, in any case for working with I guess, can anyone recommend me the best course of action?