Quake 2 PPM WIP: Austavin

[Creamy_Goodness]

> I'm working on a new model for Quake 2 of a character named Austavin, which is the alter-ego of my 15 year old nephew Austin. The mesh and skeleton are already completed, as is the stance animation, but I still have a long, long way to go before the model is completed. The model currently weighs in at 1230 patches, but I don't know yet how that translates into polys. Here's what I managed to get done so far...


> Mesh images...








> Solid images...








> Idle Spin animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Idle_Spin.avi


> Stance animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Standing_Stance.avi


> Austavin as drawn by Austin...




> Austin's Austavin YouTube channel...

https://www.youtube.com/channel/UCT5yhkvu1HU5CKVT82_IXFA


> Any feedback is welcome.

[|iR|Focalor]

No one else is probably gonna jump in with any info. I've never made animated models in Q2, but I've made static ones. A few tips...

Dunno if you've ever modeled anything for Quake2 before, but if that model with textures on it in the pics isn't already exported to and saved as MD2, you might find some rampant shifting of vertices all over the model which will create holes and gaps and distortions. Quake2 in a roundabout kinda way works on a "grid" where vertices can only be occupying certain points. So if your modeling program has vertice points placed in areas between the spots on the invisible "grid", when you convert it to MD2 format, it will shift those vertices to the nearest location that IS on that "grid" that Quake2 operates within. And THIS... is why if you try to get too elaborate with your models, they will end up looking like someone beat the living shit out of them with a hammer once you convert it to MD2. So a helpful hint... less is better. Try simplify the model as much as possible. The fewer vertices and triangles, the better. Also, might wanna seek out some advice from someone like Jay Dolan on this, he seems way more knowledgeable about the inner-workings of Quake2 itself than me, but there "might" be a way to set up your modeling program to automatically place all vertices you draw within a grid that exactly mirrors the way Quake2 and the MD2 format does it. Just a thought, dunno if it'll work.

Also, it appears as though you've taken the pictures of the textured model from a modeling program. The texture is probably nothing but solid colors, and it's the modeling program that is shading certain faces differently to create shadowing and highlights. Know that if you use a solid colored texture on the model inside of Quake2, there will be almost no shading on it. You'll have to probably freehand paint shit with Photoshop or Paint Shop or whatever to create shading and highlights.

For example, for my last pak I threw together late last year, I wanted to reskin some of the regular powerup models with high resolution skins. I didn't want to make new custom models, I wanted to them to look the same as the stock ones. So rather than creating a brand new model, I just chose to completely remap the texture to the stock models in a way that would allow me to paint a new one in photoshop easily. Was not an easy task, but I got it done.

Here's what the texture looks like on one particular model in-game.



And here's what the actual skin texture for the model looks like. Pardon the HUGE size, but... it's supposed to be hi-res, so large is better.



So what you'll have to do to create realism is freehand paint to create shading and highlighting. Also, as you can see from the texture, you can also use your freehand painting skills and some of the bevel/emboss effects within Photoshop/PaintShop to create intricate details like I did with the black rubber strap going across the back of the mask as well as the vent holes I put on the inside of the muzzle portion of the mask. This will allow you to make the wireframe model even more simple and then "artificially" make it more detailed and intricate.

One more tip, once you finish the skin texture, save it. Then you'll probably have to use some adjustment settings in the image editing program to decrease the brightness of the texture. And keep in mind that while you're doing that, any fine details you may have in darker portions of the image may appear to fade to complete black, but once you stick it in the proper folder and load it up in Quake2, it'll still appear just fine IF you don't fuck up and decrease the brightness of the image too much. If you forget to decrease the brightness of the image, then the texture will appear WAY WAY too bright inside the game, especially with brighter colors. Something you colored yellow might appear completely white. I think the settings I tweaked in Photoshop was a combination of the "image>adjustments>brightness/contrast" setting as well as slightly lowering the "lightness" value within the "image>adjustments>hue/saturation" setting. Additionally, I think I also might've adjusted the "saturation" up slightly to make some of the colors a little more vibrant after darkening it so much.

And the painting part of the project is why it's important to lay out your texture mapping properly and intelligently to allow you to paint and shade easier. Any parts that are going to be shaded using a freehand tool should be laid out on the texture map connected together on all parts where the shading will continue. Otherwise, you might see ugly parts on the finished product in-game where the shading on one triangle area goes slightly higher on the model on the neighboring triangle.

Maybe this'll make it easier, I dunno. But what I did was lay out all my triangles on the 2D texture map before painting anything, then I made a screenshot of the 2D texture map with the triangles on it at nearly the same size as the actual texture was going to be. Then I took that screenshot image in photoshop, cropped out everything but the texture map area, then I deleted all the solid color background and left nothing but the wireframe in bright green. Then I resized that to the exact size of the texture. And THEN... then I started painting. Doing this will allow you to use that wireframe as a layer within Photoshop and use it as kind of a template so you'll know exactly where all the edges of everything is located. Then you can just turn off that wireframe template layer before saving it as the finished texture.

[Creamy_Goodness]

@|iR|Focalor

> First of all, thanks for the long and detailed reply. Austavin isn't the first model I've made for Quake 2, he's actually going to be the fourth. In fact, I've actually written a currently unfinished tutorial explaining the process I go through when creating my models, which can be found at...

http://creamyg.jdluke.com/Games/Quake2/Model_Tut%28fs%29.html

> The program I used to build the mesh, and the one I'm using to animate the model is Animation: Master from Hash Inc., and it's a spline based program rather than a polygon based one. As far as the vert shifting problem you mentioned, I've never had that problem with my previous three models, so I doubt it will be an issue this time either. Austavin currently has 1230 patches, which is quite a lot, but it's still not too late for me to cut that down, if I can find a way to do so. Thanks again for all the advice.

[|iR|Focalor]

I'm a little surprised I even remembered enough about modeling to tell you that much. It's been almost a year since I've done any of it.

Your method is probably much faster. I think I was using a cracked copy of MilkShape 1.8.2 for actually drawing/designing models. I have no idea what spline versus polygon really entails. Basically I'd have to lay out vertice points on one of the 2D views, then I think I'd switch modes to "face" mode or whatever and then I'd have to click on a succession of 3 vertices points either in clockwise or counter-clockwise order to set them as a polygon/triangle and designate which way they'd face. It can probably be pretty confusing to some people because many times you'll have to switch into vertice mode, then select the proper vertices in the proper 2D view window, and then HIDE those vertices, especially when you have 2 or more points that appear on top of one another within the 2D window where you'll have to start selecting points in face mode to connect them to become polygons. And the more complex the model, the more of a puzzle it becomes to figure out which ones to HIDE and which ones to leave visible.

After modeling and exporting it from MilkShape to a MD2 model, then I'd use a program called NPherno's Skin Tool to map out the texture triangle by triangle on a blank white 2D texture image. That was a little tricky sometimes. Because you can stretch the triangles as long you want, BUT, if you do, you gotta realize that it can also end up stretching out or scrunching down the texture once you paint it. So you'll also want the triangles to be sized on the texture mapping in proportion to how they are sized on the model itself. And what I ended up doing several times was taking a sheet of blank computer paper, placing it on my monitor while the wireframe model was loaded in the modeling program, and then tracing the polygons onto the paper in pencil, then using scissors to cut it out, then placing those paper templates on the screen when I was arranging the triangles in the skin mapping tool. Sounds fucking retarded as hell and tedious, and it was, but it worked. So yeah, I wouldn't completely recommend MY methods for making Q2 models. God knows there's gotta be easier ways out there.

Looks kinda like your modeling program is drawing most of everything in rectangles rather than triangles. So that already looks way simpler to deal with than triangles. However, once the model gets converted from that program to MD2 format, there WILL be a bunch of triangles rather than rectangles. And again, the simpler the model, the better. The fewest vertice points possible is best. Also, MD2 models have a limit within Q2 as to how many total polygons they can have. And when you get started creating something, you'll say to yourself, "There's no way I'll use that many." But a lot of times, you'll find yourself getting pretty close to that number or going over it without realizing it.

[Creamy_Goodness]

> Here's a quick update on the progress of my Austavin Quake 2 model. The Run and Standing Attack animations are now completed.


> Run animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Run.avi


> Attack animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Standing_Attack.avi


> That is all.

[Creamy_Goodness]

> OK, next update. The three Standing Pain animations and the Jump animation are now completed.


> Pain animations...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Standing_Pains.avi


> Jump animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Jump.avi


> Expect the next update very soon.

[Creamy_Goodness]

> I wasn't kidding when I said that the next update would come very soon, so here it is. This is a preview of the five now completed gesture animations...


> Flipoff animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Flipoff.avi


> Salute animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Salute.avi


> Taunt animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Taunt.avi


> Grenade Toss animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Grenade_Toss.avi


> Point animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Point.avi


> As always, any feedback is welcome.

[Creamy_Goodness]

> Only a couple of animation phase updates left. This time around, I've got the five crouching animations to show...


> Crouching stance animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Crouching_Stance.avi


> Crouch walk animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Crouch_Walk.avi


> Crouching attack animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Crouching_Attack.avi


> Crouching pain animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Crouching_Pain.avi


> Crouching death animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Crouching_Death.avi


> That's all for now.

[Creamy_Goodness]

> Okay, first update of the year. The last of Austavin's 20 animations, the three Standing Deaths, are now completed. Here's what they will to look like when the model is released...


> Standing death 1 animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Standing_Death01.avi


> Standing death 2 animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Standing_Death02.avi


> Standing death 3 animation...

http://creamyg.jdluke.com/Projects/Character_Models/Austin_Luke/Movies/Animation_Tests/Austavin_-_Standing_Death03.avi


----------

> Also, I've taken some time to reduce the model's ridiculously high patch count. Before I began, Austavin weighed in at a whopping 1230 patches, but I brought that patch count as far down as I could...


> I began by turning Austavin's right hand into a permanent fist, since he uses that hand primarily to hold his weapon. I deleted the right hand, copied the right fist from my Dib model, and pasted it onto the Austavin model. This was by far the most difficult change, since I had to then make sure the fist was the right size and shape, and position it so Austavin would still look like he was holding his weapon convincingly without having to change the position of the right palm's bone. As a result of this change, the patch count was reduced by a whopping 149 patches, bringing the model down to 1081 patches.






> Once the fist was done, I moved on to the ears. By flattening out the concave front of the ear, I chopped off 13 patches from each ear, totalling a reduction of 26 patches. This further lowered the overall patch count from 1081 to 1055.






> The next part of the model to be edited was the bottom of the shirt. I removed the middle spline ring and reconnecting the rest of the splines, and this lowered the patch count by another twelve, bringing the patch count down from 1055 to 1043.






> After editing the bottom of the shirt, I moved on to the irises of Austavin's eyes. Since the back six patches were completely hidden by the eye whites at all times, I chopped out those patches. I also flattened out the front of the irises by deleting the center vertces and replacing them with horizontal splines, cutting four additional patches from each iris. I ended up chopping a total of 20 patches from the eyes, ten from each iris, and this lowered the model's patch count from 1043 to 1023.






> The lips were simple and had only three cross sections. The back cross section was completely hidden by Austavin's face at all times, so I deleted all five back splines that made up that cross section. This brought the model's patch count down by another nine patches, reducing the total patch count from 1023 to 1014.






> Next up to be edited was Austavin's head hair. After I deleted a few splines from the front of the hair, the hair was lightened by six patches, bringing the total patch count down from 1014 to 1008.






> After the head hair came the eyebrows. I deleted three permanently hidden patches from the back of each eyebrow, totalling six patches taken off of both eyebrows. This reduced the model's patch count from 1008 to 1002.






> While working on the reduction of Austavin's patches, I discovered a hole on either side of the head on the hairline between the ear and the eye. I fixed these holes by adding a single polygon to either side of the head. This fix added two patches to the model's total patch count, increasing it from 1002 to 1004.










> At this point, the animation phase is completed apart from the 198 frame animation choreography, which I plan to work on today. Once that's done, the animation phase of this project will be officially completed, and then it's on to the optimization phase. After I get through that, I will be writing the next update to this page.

[Creamy_Goodness]

> Ten days ago, I began the optimization phase of Austavin's Quake 2 PPM. I exported the first frame of Austavin's All Frames choreography as 3ds file and imported it into Quake Model Editor, where I used the standard Male Quake 2 model as a reference to correctly adjust the rotation, position, and scale of Austavin. Then, I exported all 198 of Austavin's frames as a series of 3ds files, and then saved them as a single md2 file in QME. The polycount was unacceptably high, weighing in at a whopping 1714 polygons. At this point, the model was a mess, since half the polygons were facing the wrong way, so I flipped the errant ones on body part at a time. As I was working on the hair, I noticed gaps where the hair met the forehead, so I went back into Animation: Master and buried the hairline's verts into the forehead. I also took this opportunity to remove even more unnecessary patches from the model...

> The top of the crotch was permanently and completely buried within the bottom of the shirt and would never be seen. There were four patches sealing this area closed, and after removing them, Austavin's patch count went down from 1004 patches to an even 1000.


> Patch Reduction - Crotch - Before Fix...




> Patch Reduction - Crotch - After Fix...




> The tops of the shoes were completely buried within the bottom of the pant legs. There were two patches sealing the top of each shoe closed, and after I deleted these four patches, the patch count went down from 1000 to 996.


> Patch Reduction - Shoes - Before Fix...




> Patch Reduction - Shoes - After Fix...




> I saw what I thought was a chance to shave off a lot more patches from the model when I examined the back of Austavin's shirt. I saw that this part of the model was somewhat flat, and the central vertical spline may not be needed, so I removed it and adjusted the patches above and below it in order to accomodate the change. Surprisingly, for all the work I put into this edit, I only managed to lower the patch count by two, bringing the count down from 996 to 994.


> Patch Reduction - Shirt Back - Before Fix...




> Patch Reduction - Shirt Back - After Fix...




> After going through the exporting process again, I found that Austavin's polycount had dropped from 1714 to 1690. I then went on to correctly flip the rest of the normals in the model, but I also had to apply backfaces to the hair, and this increased the polycount to 1752.

----------

> Once all of Austavin's polygons were facing the right way, I moved onto eliminating the model's permanently hidden, and thus useless polys. I began with the hair, where I found a total of four hidden polys near the ears, two on each side. After deleting them, the polycount dropped from 1752 to 1748.


> Hidden Poys - Hair 1...




> Hidden Poys - Hair 2...




> Within the ears, I found six hidden polygons in the middle of each ear, totalling twelve. Deleting these polys lowered the polycount from 1748 to 1736.


> Hidden Poys - Ears 1...




> Hidden Poys - Ears 2...




> Next up was the eyebrows, where I found two hidden polys inside each brow. Getting rid of these four polys further lowered the polycount from 1736 to 1732.


> Hidden Poys - Eyebrows 1...




> Hidden Poys - Eyebrows 2...




> The open mouth segment had only 40 polygons total, but eight of them, a fifth of the total, were hidden and useless. Once they were gone, the polycount dropped from 1732 to 1724.


> Hidden Poys - Open Mouth 1...




> Hidden Poys - Open Mouth 2...




> After all of the extra polygons in the head had been eliminated, the next step was to drastically edit the chest area in a way that would take down the polycount dramatically. The collar of Austavin's shirt was open so that the polys of the chest could be seen. My plan was to seal off the collar, eliminate the chest under the shirt, and later to simply draw the collar onto the skins during the upcoming skin phase. Sealing the collar closed added nine polys to the model, increasing the polycount from 1724 to 1731.


> Collar Seal - Before Fix (Wireframe)...




> Collar Seal - Before Fix (Shaded)...




> Collar Seal - After Fix (Wireframe)...




> Collar Seal - After Fix (Shaded)...




> Once the shirt collar was sealed, the polys in the chest were now completely hidden and useless, allowing me to delete the entire chest. This wiped out a whopping 44 polygons from the model, bringing the polycount down from 1731 to 1687.


> Hidden Poys - Chest 1...




> Hidden Poys - Chest 2...




> With the chest now gone, I moved on to fixing the hands. I was able to find 13 hidden polys buried in the five fingered left hand, most of them at the base of the fingers. Removing them lowered the polycount from 1687 to 1674.


> Hidden Poys - Left Hand 1...




> Hidden Poys - Left Hand 2...




> As for the right hand, which was a permanent fist, I was able to find ten hidden polygons. After I deleted them, the polycount dropped from 1674 to 1664.


> Hidden Poys - Right Hand 1...




> Hidden Poys - Right Hand 2...




> When I examined the crotch segment for hidden polys, I managed to find six of them near the bottom. Once they were gone, the polycount went down from 1664 to 1658.


> Hidden Poys - Crotch 1...




> Hidden Poys - Crotch 2...




> The next section to be fixed was the shoes, and this area was particularly bad. There were a dozen hidden polygons in each shoe in the form of a ring around the top section. These polys were actually inside 24 poly rings, and shared the same vertices as them. I simply exposed the 12 poly rings, and eliminated them, lowering the total polycount from 1658 to 1634.


> Hidden Poys - Shoes 1...




> Hidden Poys - Shoes 2...




> The final section I worked on was the bottom of the shirt. There were 20 polygons sealing the bottom edge of the shirt, and I felt that I could seal that edge with fewer polys. I eliminated the middle row of vertices and reconnected the outer ring of verts with a new set of just ten polygons. This change made the bottom edge of the shirt flatter, but also reduced Austavin's polycount from 1634 to 1624.


> Poly Reduction - Shirt Bottom 1...




> Poly Reduction - Shirt Bottom 2...




> Poly Reduction - Shirt Bottom 3...




> With all of the visible polygons facing the right way and the errant ones eliminated, the optimization phase of the Austavin Quake 2 PPM project is now completed. The next phase will be the skin phase, and I'll update this log again once that phase is over.

[Creamy_Goodness]

> While working on mapping the polygons of Austavin's hands in Npherno's Skin Tool, I found a strange indent on the base of the left thumb. I was able to fix that by opening the model in Quake Model Editor, deleting the three polys and the isolated vert that made up the indent, and then patching the hole with two new polys. This lowered the overall polycount by one. I also found two additional hidden polygons inside the first joint of the thumb that I had somehow missed during the optimization phase, so I deleted them. After these fixes on the thumb were finished, Austavin's overall polycount dropped from 1624 to 1621.

> Later on in the skin mapping process, I found that the back two polygons in the top row of the crotch segment would never be visible during gameplay, and so I deleted them in QME. This brought the polycount down from 1621 to 1619. At this point, I don't think I can lower Austavin's polycount any further. The skin map took several days to complete, but when I did, it turned out to be 256 x 256 pixels in resolution with 17 percent wasted space. The next step is to draw the default skin, a task that I plan on starting tonight.

[Creamy_Goodness]

> Austavin's default skin is now completed, as well as his two Capture The Flag skins; red and blue. This concludes the skin phase of the Austavin Quake 2 PPM, and I will soon be moving on to the fifth and final phase, the visible weapons package, or VWEP for short.


> Mesh...




> Flat shaded...




> Austavin skin...








> CTF Red skin...








> CTF Blue skin...

[Creamy_Goodness]

> The fifth and final phase of my Austavin Quake 2 PPM is currently underway. This phase is the VWEP phase, which stands for visible weapons package, and involves creating weapon models for Austavin that will represent each of the twelve weapons in Quake 2, including the grappling hook. Austavin will use my universal 1-handed VWEP, which he will share with two of my previous Quake 2 PPMs, Chuckie Finster and Dib. However, he will use a unique hand grenade model, which is basically a pipe bomb with a wooden handle. The hand grenade model is completely finished; the model has been built in Animation: Master, optimized in Quake Model Editor, skinned in NPherno's Skin Tool, and then scaled and aligned to the right hand of the Austavin model in QME. Here's a few preview pics that will give you an idea of what the hand grenade will look like...


> Mesh...




> Flat shaded...




> Skin...






> Three of Austavin's other weapons; the blaster, the railgun, and the grenade launcher, are all pre-existing models for the universal VWEP, so all I had to do was to align them to Austavin's hand. The other eight weapons will be aligned to Austavin's hand over the next few days, and then I will try to officially publish the first official release version of the PPM on the 20th of this month.

[Creamy_Goodness]

> At long last, after about 10 months of on and off work, I am finally ready to release the first official version (Version 1.1) of my Austavin Quake 2 PPM. The model is far from perfect and there are many things about it that I'm not happy with. First off, at 1619 polygons, the polycount is uncomfortably high. Second, there are weird lines in the skin that only seem to show up in OpenGL mode. Third, I was forced to place the weapon model unusually high up in Austavin's right hand in order to prevent it from clipping too badly into the left hand during the run animation. However, I can fix all of these issues in future versions of the model, if I ever get around to it. But for now, I'm all burnt out and I consider the model to be good enough for the time being. My nephew Austin will be using this model a lot when he plays on Tastyspleen, so if you want to see him properly, go grab yourself a copy of it.

http://creamyg.jdluke.com/Projects/Quake2/PPMs/Austavin/Austavin(v1.1).zip

> Enjoy.

[Creamy_Goodness]

> Merely 18 days after the initial release of my Austavin Quake 2 PPM, I'm already ready to release a minor update of the model. In the initial release, I had forgotten to switch the color mode of the Capture The Flag icons from RGB to Index Color, and to apply the Quake 2 palette to the icons. As a result, attempting to use the CTF skins resulted in an error which made the skins unplayable. This problem has been fixed in Version 1.2. The model also had a minor issue where thick seams would be visible on certain parts of the model while running in OpenGL mode. By tweaking both the skin map and the skins themselves, I was able to eliminate the worst of these lines, particularly on the sides of the shirt and on the arms. However, this fix increased the Wasted Space of the skin map from 17% to 20%, but I think that making the model look better in game was well worth it.

http://creamyg.jdluke.com/Projects/Quake2/PPMs/Austavin/Austavin(v1.2).zip