Author Topic: enemy models  (Read 25844 times)

omega res novae

  • *
  • Posts: 541
  • previously known as squall8cloud7
    • View Profile
    • omegas mods google docs
Re: enemy models
« Reply #25 on: 2012-10-25 14:22:04 »
you are legit whiteraven. ill start porting today if i get the chance. ill see if the the 2 ?'s in the op can use the same model. if yes then ill do those as well. if not ill adjust the model. they cant be that different

BlitzNCS

  • *
  • Posts: 889
  • Master of nothing in particular
    • View Profile
    • My Youtube
Re: enemy models
« Reply #26 on: 2012-10-25 14:55:58 »
I can't speak for Timu, but I'd say you'd need to make some changes if you want this to be usable.

Looking at your wireframe, I'd say you really just need to look at optimising your model.

As great as it is to have mega high-res graphics crammed into as many parts of the game as you can, realistically, it's not necessary to have what looks like a 36 sided cylinder for each lower leg. This applies to the texture as well - you should only really be at a max of 5122 in this situation. It'll be a lot easier to work with at a smaller size, too.

Basically, just reduce the amount of polygons as far as possible until you actually start losing detail at the maximum zoom level used by the game itself - you don't need any more detail than that because nobody will ever see it, and also because you're making the graphics renderer do more work even though you're not seeing any noticeable difference.

I'm not saying you're going to slow down people's systems, and theoretically you can have as many polygons as you'd like, but a few of those on screen at the same time as well as all the other high resolution battle models and textures and you might start pushing some slightly older machines.

Optimisation is just a good habit to get into when working with models, especially when you're doing stuff for video games.
If it were me, I'd probably use max 6 sides for the limbs (not counting the thighs), and I'd probably try and at least half the polygons for the body and head. Remember, it's such a small model, and you rarely get to see it very close up.

Good Luck!
« Last Edit: 2012-10-25 14:57:50 by BlitzNCS »

omega res novae

  • *
  • Posts: 541
  • previously known as squall8cloud7
    • View Profile
    • omegas mods google docs
Re: enemy models
« Reply #27 on: 2012-10-25 15:45:41 »
i understand what you mean. ill minimize the polys where i can. timu commented on a thread before. 12 sides around is what he said then. for caits weapons. ill optimize where i can but that may just be the feet and the limbs.

Mayo Master

  • Moderator
  • *
  • Posts: 650
    • View Profile
Re: enemy models
« Reply #28 on: 2012-10-25 16:07:12 »
Regardless of NCS's comments, I have a pet peeve about the frog model, which concerns the mouth  :P (omega knows about this). Namely, a frog's mouth isn't flat, but parabolic, from below one eye to the other. I feel weird to be the only one to point that out!

Looking at your wireframe, I'd say you really just need to look at optimising your model.
[...]
Even though you have much more knowledge than me about what's good practice in the video game industry, I have the following questions/comments (it may sound naive, but what the heck):
- Wouldn't it be good practice to make a very high-poly model, and then make downgrades from it which would in turn be incorporated in the game?
- I also wonder if the aforementioned suggestion would be best in our situation, given that our project has no "release date". I remember someone speculating in a release circa 2030, and that may be more realistic than not. In this case, low-poly models can quickly become outdated (a game like Gran Turismo 5 had this kind of issues). If one stores a very high-poly model somewhere, this one would be somewhat "time-proof". The downgrades from the high poly model could be performed depending on the computing power of the generation of hardware when they're released. Besides, I believe that most 3d modelling software (Blender among them) have very simple-to-use algorithms to make such downgrades.

Maybe I'm wrong, but I thought this aspect was to be accounted for before deciding of a modelling strategy.

omega res novae

  • *
  • Posts: 541
  • previously known as squall8cloud7
    • View Profile
    • omegas mods google docs
Re: enemy models
« Reply #29 on: 2012-10-25 16:15:17 »
maybe by 2030 frogs will have a flat mouth. i modeled this around the original without looking at any reference material. to go about fixing the mouth idk how id do that

also still waiting on the oppinions of someone that can grant approval

Mayo Master

  • Moderator
  • *
  • Posts: 650
    • View Profile
Re: enemy models
« Reply #30 on: 2012-10-25 16:48:41 »
I tend to believe that the mouth was flat in the original model because it's a very low-poly model, rather than a design choice (remember the image I sent you? this one was precisely the concept art for FF7's frog). As for modifying the mouth shape... I suppose subdividing faces and moving the vertices around should do it, shouldn't it?

whitERaven

  • *
  • Posts: 399
  • when the Raven turns white..
    • View Profile
Re: enemy models
« Reply #31 on: 2012-10-25 17:17:28 »
LoL.. no offense mayo.. but I just find it wierd to worry about a flat mouth being different from the real frog's mouth while infact we have a frog here standing on two legs and has hands.  ;D ;D ;D

But then I do understand that realistic look has it's good points but I don't think were bound by it too much while designing characters, In the end what looks good is what matters.

Mayo Master

  • Moderator
  • *
  • Posts: 650
    • View Profile
Re: enemy models
« Reply #32 on: 2012-10-25 17:37:39 »
Omega had the same reply to that comment before, about how it's futile to worry about the mouth while the frog stands on its legs. I'll try to explain my pet peeve: while the frog stands on its legs, the legs shape itself makes sense (compared with a realistic picture, obviously the legs have a different scale, but that's not the point). A flat mouth shape just seems wrong, in my opinion.
But eh, it's just my opinion, and if nobody else cares, suit yourself.

whitERaven

  • *
  • Posts: 399
  • when the Raven turns white..
    • View Profile
Re: enemy models
« Reply #33 on: 2012-10-25 17:54:51 »
It's not like it's a bad idea.. like I said for me what looks good is what matters, if changing the mouth to what you wanted will make the model look better, then I think it'll be good if omega heed your advise, well it depends if omega is up to it..
That is just in my opinion too so it'll depend on what the author thinks :) ;) :D..
« Last Edit: 2012-10-25 17:57:12 by whitERaven »

omega res novae

  • *
  • Posts: 541
  • previously known as squall8cloud7
    • View Profile
    • omegas mods google docs
Re: enemy models
« Reply #34 on: 2012-10-25 18:21:20 »
Ill see what i can do. timu said head and face need more charactor. also a friend of mine thats a professional animator said it resembles a snakes head. so ill see what i can do.
for the mouth i imagine deleting some polys along the mouth and where the mouth will go
delete corners of the mouth.
extrude it to its new length. fill holes

ive thought before the cheeks were flat. this will give me a chance to round them.
eyes idk maybe round out the backs.
« Last Edit: 2012-10-25 18:49:57 by omega res novae »

BlitzNCS

  • *
  • Posts: 889
  • Master of nothing in particular
    • View Profile
    • My Youtube
Re: enemy models
« Reply #35 on: 2012-10-25 18:39:07 »
i understand what you mean. ill minimize the polys where i can. timu commented on a thread before. 12 sides around is what he said then. for caits weapons. ill optimize where i can but that may just be the feet and the limbs.

Just remember, 12 sides is not likely to look any different from 6 sides at the resolution the frogs limbs will be displayed at. Especially since the only shading of the model will be done via your texture.

This is an image of two cylinders displayed with the battle engine's shading. Can you tell the difference? One has 8 sides and the other has 36. And this is displayed at a much higher resolution than those legs will be seen at. The majority of your detail should come from your texture, as that's what the battle engine is designed to display the most detail with.

The feet are looking fine if I'm honest, I wouldn't really change those too much.

- Wouldn't it be good practice to make a very high-poly model, and then make downgrades from it which would in turn be incorporated in the game?

Not unless you're going to be using it in another situation where you'll be zooming in very close and rotating around an individual limb. 'Downgrading' isn't as simple as you may think.
It's like building a huge rocket for the sole purpose of heating some milk in your kitchen occasionally. Or something.

- I also wonder if the aforementioned suggestion would be best in our situation, given that our project has no "release date".

Considering the production of FF7 was completed in 1996, the battle engine isn't likely to suddenly spontaneously change. The only other factor to really consider is typical screen resolution, which I don't think is likely to completely skyrocket anytime soon. Besides, 1080p HD is by far enough for this kind of project. Hell, updating the graphics to look nice at 640*480 would be a vast improvement.

Regarding the mouth thing, the point of this project is to build something around the original source using artistic license and original artwork to fill in details that aren't clear. As long as you're keeping roughly true to the original model, it's basically up to the artist.

Tempus

  • *
  • Posts: 56
  • This guy are sick.
    • View Profile
    • Deviant Art
Re: enemy models
« Reply #36 on: 2012-10-26 15:45:22 »
Not unless you're going to be using it in another situation where you'll be zooming in very close and rotating around an individual limb. 'Downgrading' isn't as simple as you may think.
It's like building a huge rocket for the sole purpose of heating some milk in your kitchen occasionally. Or something.

I second this. Not that my opinion matters or anything :D

There's two purposes that I can think of for having a high and low poly version of a given model. The first, like BlitzNCS noted, is if you're using it in another situation where the extra detail will be noticed such as an ultra-zoom in or in an FMV. The second is if you plan to bake the detail from the high poly onto the low poly. FFVII doesn't support normal maps to my knowledge, but you could potentially bake ambient occlusion from the high poly and blend it into the diffuse. But you really don't even need a high poly model for that; you could just bake the AO from the low poly I doubt there'd be a noticeable difference. And this is really at the heart of the matter: is the difference noticeable?

I also agree making low poly versions from high poly is more difficult than some think. Yes, you can use something like the decimate modifier in Blender, but it can't be controlled (besides changing raw face count) and going from 100,000 tris to 500 tris can produce unpredictable and undesirable results (such as long, thin triangles that need to be manually edited later anyway). You get the best results from retopologising (i.e., creating the low poly on top of the high poly)... which also happens to be the slowest process!

Ideally, a project's workflow is decided beforehand - either high polys are needed or they aren't (or certain predetermined models will be featured in situation X where a high poly is needed). Making them when they won't be used or noticed wastes dev time.
« Last Edit: 2012-10-26 15:47:54 by Tempus »

omega res novae

  • *
  • Posts: 541
  • previously known as squall8cloud7
    • View Profile
    • omegas mods google docs
Re: enemy models
« Reply #37 on: 2012-10-30 17:00:48 »
redid arms and lower leg they are now 8 sided. i will do the feet next. i extended the mouth some. i made the uv fit whiteravens texture just to have a general idea of what itd look like here are the results. itll look better once the texture is done.



Nightmarish

  • *
  • Posts: 525
  • owari da
    • View Profile
Re: enemy models
« Reply #38 on: 2012-10-30 17:20:22 »
How does that "realistic" texture look ingame?

whitERaven

  • *
  • Posts: 399
  • when the Raven turns white..
    • View Profile
Re: enemy models
« Reply #39 on: 2012-10-31 02:50:44 »
Haven't tried it, since model is not yet ported to ff7 format.. But I suspect it won't look as good, so after model is ported I'll make adjustments in the textures to make in more or less appear like those rendered images.

Almighty_gir

  • *
  • Posts: 242
    • View Profile
Re: enemy models
« Reply #40 on: 2012-11-01 11:01:22 »
basic rule of thumb here, is think about how many pixels something is going to take up on-screen, and for how often.

from my memory, those frogs are tiny, they take up maybe 100 pixels (at the very very most). so you should probably try to put their entire texture onto a 256*256 texture map (personally i'd go 128...) and a poly limit of 150-200.

omega res novae

  • *
  • Posts: 541
  • previously known as squall8cloud7
    • View Profile
    • omegas mods google docs
Re: enemy models
« Reply #41 on: 2012-11-01 11:38:25 »
the size of a texture is trivial. as long as it keeps the same proportions they can be resized and touched up easily.

Almighty_gir

  • *
  • Posts: 242
    • View Profile
Re: enemy models
« Reply #42 on: 2012-11-01 12:25:14 »
http://www.poopinmymouth.com/tutorial/tutorial.htm

in particular: http://www.poopinmymouth.com/tutorial/resize.htm

Quote
True resolution vs. Double Size

When painting a texture, many times people will paint at two times the resolution, then size it down. Meaning if the final map is a 512x512, they will paint at 1024x1024. I recomend staying at true resolution whenever possible.

If you are just starting out at texturing, painting at the actual resolution is a good habbit to get into. There are many reasons for this. First and formost, is the time factor. You do not have to worry about any details that cannot fit into the end result map, because it IS the end result map. If the pixels hold it, it will show up in game (if the camera/player gets close enough) Secondly is how crisp your end result will be. I have said before, and Ill say again, you as the artist, want the control. any uniform application by a computer will end up looking piss poor compared to the same sort of effect applied by a skilled artist. Each pixel in the resized map is the result of the average of four pixels at the larger size. You can place single pixel highlights, as well as razor sharp single pixel seams and shadows. They will not be blurred to obscurance by photoshops algorithms.

Common reasons people argue for painting at double res:

But I can be messier/faster/scribblier and then size it down and it looks "right". This is not true, it only makes your mistakes less apparent. Their is a reason that art teachers tell you to "draw big" It is because it makes your mistakes apparent. If you cannot make it look right at the small res, you are not making it look right when you resize either. It is just that your mistakes become less apparent.

But I can get subpixel detail! Again, not true, there is no such thing. A pixel is a pixel is a pixel. You can have detail that is implied as smaller than a pixel, but it is just as reproducable at res, as it is double size, you just have to know how to do it. and it will read better because you did it at true res, than if you resized down. One particular thing that people like resizing for is hair. A good way to learn how to do implied sub pixel detail, is to actually resize something, and look to see what the pixels are doing that implies that thinner than a pixel detail. Once you learn these "rules" you can use them or adapt them for your needs, and be faster because you are painting them at res.

Again, if you are already fantastic at texturing, and you resize. By all means, continue doing it. It is, and always will be, the ingame results that count.

whitERaven

  • *
  • Posts: 399
  • when the Raven turns white..
    • View Profile
Re: enemy models
« Reply #43 on: 2012-11-05 02:11:47 »
Yep I'm one of those who paint on double res, then resize in the end.. somehow it's easier for me to do brush strokes in bigger res.. It's always messy applying/using different brushes in small res images..


PitBrat

  • *
  • Posts: 1376
  • Change to feed the machine.
    • View Profile
Re: enemy models
« Reply #44 on: 2012-11-05 02:14:59 »
Doesn't that adversely affect the integrity of the edges?

whitERaven

  • *
  • Posts: 399
  • when the Raven turns white..
    • View Profile
Re: enemy models
« Reply #45 on: 2012-11-05 02:22:43 »
Sorry pit, I don't really get what you mean there, LoL I can be dumb most of the time, ;D ;D

Almighty_gir

  • *
  • Posts: 242
    • View Profile
Re: enemy models
« Reply #46 on: 2012-11-05 13:12:18 »
what he means, is that if you have a 1 pixel line which defines an edge at 256*256, when you resize down to 128*128 (which is 1/4 of the size), what are you going to see happen to that 1 pixel line? it can't magically turn into 1/4 of a pixel thick. instead it gets interpolated with the pixels surrounding it, which makes it less apparent.

whitERaven

  • *
  • Posts: 399
  • when the Raven turns white..
    • View Profile
Re: enemy models
« Reply #47 on: 2012-11-06 02:22:10 »
Ah.. I see,, but most of the textures which I used the double res methods are those where I need to paint faces which involves a lot of smudging so there not much defined edges involved, maybe that's why I did not notice issue..

Mayo Master

  • Moderator
  • *
  • Posts: 650
    • View Profile
Re: enemy models
« Reply #48 on: 2012-11-06 15:16:15 »
Then I have a noob question:

Would downscaling resolution (and having edges less apparent) actually give a benefit when the object you wish to model is meant to have rounded/soft edges?

Almighty_gir

  • *
  • Posts: 242
    • View Profile
Re: enemy models
« Reply #49 on: 2012-11-06 19:29:23 »
not really, all downscaling really does is blur your image... here's a quick example:

you have a 3*3 image (9 pixels), going clockwise from the top left pixel you have the following values:
255, 0, 0
0, 255, 0
0, 0, 255
255, 255, 0
255, 0, 255
0, 255, 255
0, 0, 0
126, 126, 126

and in the centre: 255, 255, 255.

this should show that you have; red, green, blue, magenta, cyan, yellow, black, gray, white.  essentially all of the "extremes" of the colour spectrum.

now resize that down to a 1 pixel image, and you will get a single gray pixel. in fact you can order the colours in any way you like, and you will always get gray when you resize. what this should tell you, is that when you resize your image down, you lose information. it should be self explanatory... you can't shrink an image (therefor making the pixel density smaller) and expect to keep all the detail (which would require all of those pixels to remain in the image). but many people stallwartly believe that they "do better" by resizing... it's just an illusion really, and as the article i linked above tells you, it just hides your mistakes.

so, if you think resizing helps you make better art, i would argue that you should practice at true res, because it will make you a better artist, and therefor you will make better art.