MTM2.com

Okay, one thing leads to another and it's time to beef up my truck painting. Some of you guys are doing amazing work and I'm feeling left behind, lol. Time to step it up to a more advanced level.

Now, we all know that phototexturing can look great but it can also look very shoddy if either the source image is blurry, or doesn't reduce well, or contains too many colors and the like. And the methods have been explained well enough by Alpine in his <a href="http://cownap.com/~mtmg/trucks/painting2.html">Star Craft</a> tutorial. So, sure, I've played around with phototexturing and had mixed results - some very good, some very very bad.

I have a hundred questions, but the first I'd like to concentrate on is the idea of hand painting. Most every truck that has been hand painted looks excellent. So, question number one is: When you say you hand paint, what precisely do you mean? Do you have one of those scratch pads that draw right into the comp, do you manage with just the mouse, or do you use some other method, or what?

Thanks.

As for hand painting, I mean I paint all my trucks free hand, and with my mouse. Imagine you have one solid white square, I trim it down with a black outline, till it fits the shape of the panel. Then I just use the line thing in PSP and draw lines till it ends up purdy. When I do a rep, I set the picture of the truck as my wallpaper, and start painting. Minimizing and maximizing, till I get it right.

I've never cut & pasted anything from a picture ever. So when I say free-hand, that's what I mean, I really do free hand it. There may be an easier way, but I love doing it the way I do. I got so sick of seeing trucks that were so cool, but just looked like crap becuz they cut & pasted the logos from a picture, what a waste.

So when I learned how to make trucks, I wanted my paintjobs as crisp and clear as I could get them. Then for different shades and simulated body lines, I just use those white boxes in PSP when you change colors at the bottom. Once I get the color exact, I save it in one of those boxes. Then do the same to 2 more, one darker, and one lighter, then painted in the right places, it makes a totally flat panel look like it has molded in bodylines.

Then once I get one section done, I save the paintfile, but leave PSP open. Then I conevert it back, and look at it on the body in Binedit. I line everything up to the exact pixel for the body and the box sides, then everything is straight and neat. If I'm off, I just adjust it in PSP and keep saving it over and over, and constantly checking the alignment in Binedit, then once one side is done, I copy & paste those sides to the other side, and do the same by lining everything up. So if you look at one of my trucks, there will be a line that starts at the grille, and goes straight down the side, and lines up exactly to and around the tailgate, and then around and down the other side. It's alot of work, but when you're all done, it almost looks real. That's the part I love.

Hand painting to me is just anything not phototextured. RF does the pixel by pixel thing with great results, I prefer to use PSP 7's fantastic tools to make something anti-aliased yet still crisp and clear. I do have a tablet which makes painting 2000 times easier (If you look at OutRageous, the rips wouldn't be possible without the tablet, as would the lettering).

<center><img src="http://cownap.com/~trucks/P2004/outrageousbycale.jpg" width="400"></center>

I've been meaning to somehow catalog all of my PSP tricks for truck painting. Using layers, anti-aliasing, gradients, filters and effects, picture tubes, etc. When I get to school next week maybe I'll whip up a webpage or something on it.

Now mapping is something I'm still learning, but at Betrayal can attest thanks to a truck I mapped for him (not released yet) I am learning quickly

Thanks for the quick replies.

Cale, when you say 'rips' do you mean those flamey type things at the top of the pinkish part that goes up into the yellow upper half? And by the way, if you come up with a tips page, I'll host it up in the truck section of the mtmg if you need a good spot for it.

Rep, thanks. Gives me a ton to think about and work with. I'll try some of that out and see what happens. No tablets here yet for me =(

Okay, so the next question up is probably one best left to drive2survive, but I welcome alternate ideas too, of course.

On this page --> http://mtm2.com/~d2s/binedit/tute/bined_tute3d.html

Down near the bottom, d2s describes making fitted textures from scratch. Now, I've played with this and I've played with this, but never never never can I get it to come out like his falcon textures. Can somebody expand on those notes or make extra suggestions. When I try, the mapping window dots get in the way, the shape is always wrong, and it ends up being pretty much useless - which can get somewhat frustrating, let me tell ya, hehe.

The "rips" are the pink splashes. They were hand drawn by scribbling on the tablet, then I used the Inner Bevel effect to make them 3 dimensional (after selecting them with the magic wand).

I'll take you up on the tips page offer.

The method D2S used there is something I used to map the truck for Betrayal. It's a complicated 3d sculptured character body truck (you know, like Snake Bite and Bulldozer), with very complicated curves. Using that technique made what was otherwise impossible look incredible.

What I did with that is to select a portion of the body as a face group (trying to stay as much on the same plane as possible lest BE messes with it), save it, then map it to a black texture. I make sure that the mapping window is as close to 1:1 zoom as possible, print screen it to the clipboard, then paste it to PSP. After making sure that there are layers I will paint on underneath it, I then select all the black spots in the image, delete them for transparency. That leaves an outline that you can paint under. When you are done, simply delete the layer with the poly outline, load the .raw into BE and map to the face group (if you saved the model after mapping the face group to black, you should be able to just use replace texture). Should be almost perfect.

Ahh, grasshopper... I am giddy with the role reversal LOL.

I'll do this as a two part post... I'll just start off briefly with my take on hand-painting. Like Rep, I use the mouse, and I mostly now use PSP7 over PSP4 for its improved tools and better colour reduction function.

Now in some cases if I have a semi-decent photo image, I'll actually phototexture the truck and use the photo image as a guide to paint over. Example:



When I did that one I had a good side-on photo to work with, and I was repainting one of Karm's bodies so I already some texture files laid out for me. For the large and fancy "Wild Thing" lettering on the side, I copied the from the photo text onto my truck textures , sized it to fit, then I selected some primary colours and drew over the lettering to make it crisper and remove the photographic artifacts and colour variance (since it typically doesn't reduce to 256 colours very well).




If it's a case where I don't have good photo(s) to use for tracing over, as with this recent work, then I do like Rep - open the photos somewhere I can see them and draw construction lines onto the textures referencing from the photo, and periodically generating RAW files from them and checking the results in BinEdit. When I have the shapes and markers in the right places, I fill in the colours and detail proper. I also hunt around online a lot for graphics I can apply straight to the truck - things like a Ford badge or a TripleM logo can be quite easily found and it's far easier than drawing those by hand too. To finish I use the airbrush tool and opacity settings to antialias my drawn edges, brushing over the curved lines to get rid of the jaggies. Nowadays I always save my work-in-progress textures in a truecolour format (BMP or PSP file) to keep the colour depth right up until I'm ready to make my final RAW textures, so I'm not converting back and forth between 256 and true-colour and messing with the colours of the image.

Well, that all was just something I made up at the time - in making an entirely new model, obviously I needed to make an entirely new texture set to go with it and I was trying to find the easiest starting point. Basically all it boils down to is taking a screenshot of the model mesh, pasting it as a texture and then painting over it - like my tracing over the top of photographs above.

You can take screenshots and copy straight from BinEdit's orthographic display, although I find that other parts of the model can interfere (though you can temporarily delete them) and scaling to fit within a 256x256 RAW can sometimes be a problem. The idea with the mapping window is just to select the faces you're interested in, let BinEdit lay the mesh out within the 256x256 square for you, and just scale it to whatever size you want as if you were mapping. Then take a screenshot (use a 1:1 zoom in BE2's BinMapper) and copy and paste the mesh onto 256x256 image for your texture.



That pic's a bit dumb since it's a red background blending with the red vertex markers, you should use something that contrasts like black or white. Anyway, once you have the mesh outline you can fill in the interior with some base colour of your choosing. You don't have to keep the lines from the mesh - I choose to leave them there since they made a good guide for mapping the model proper (but isn't really necessary if you all you need to do is move and scale the mesh), and also because the lines I built into the model corresponded directly to details (windows, panel joins etc) I wanted to paint onto the textures.



I my case I chose to fill the interior of the mesh with red, and the lines with a brighter red. Use PSP's colour-replacer tool to replace the lines and dots you got from the screenshot of the BinMapper with the colour you want (and make them all the same colour) with a simple double-click. At this stage I mapped the textures to my model so I could use the wireframe on the texture to line up the mesh in the mapper, but as I said that's really a bit of overkill - you just need to make the mesh match onto your texture shape.

Then you just paint it however you want (if you've pre-mapped it, then you can keep checking in BE how the thing looks).



You might find that BinEdit will not lay the mesh out for you the way you want it in the mapping window. The areas you want to map must be reasonably flat otherwise BE may distort them or lay them out on the wrong orientation. For example, when I was building that falcon, I wanted to texture and map the front bumper as one piece. However, the bumper curves around the front of the car so it doesn't really appear flat from any angle. When I selected the faces and invoked the texture mapper, BE laid the wireframe out as viewed from above - which was completely useless. I ended up using screenshots of the front and side orthogonal views, cutting out the bumper and piecing fragments from the front and side together to make an approximation of the bumper laid out flat. Then I had to hand-map it by painstakingly moving each vertex onto my drawn mesh.



Clear as mud?

> Clear as mud?

Well, it fills in some missing steps. I'll try my hand at a few things and let you know.

Cale, send me an email when you're ready.

Thanks all. Give me new things to play around with :-)

In a way, I do what D2S does. I'll use Radical Rescue for an example. First thing I did before I even thought about the paint was model the body. Once I finished that, I (actually did it the hard way, lol) cut out the individual panels, and took a screenshot of the wireframe.

Over to PSP... Paste it, crop it, fill in the wireframe, and blackout the rest. From here, I applied the shading to the body. I approximated where the lines should be, and applied greyscale fades in the proper places. If done right, the finished shading should look like a rolled piece of sheetmetal.

I then do something that very few other authors do. I add a layer. On this layer, I apply the base paint. Add another layer.... This one contains all the decals logos, etc. Finally, add another layer and paint in your windows. Then just set the base paint to an overlay, and fiddle with the rest until you get the desired effect. But always leave the windows as normal...

And once it's all said and done, you get this.



I also used this process to paint my now nonexistant Carolina Crusher replica. Gotta love all the evils of computers.

Yes, I use layers all the time. Layers and saved selections can really cut down on the time and labor. I imagine if you're painting over a photo, that using the slider on the layer to fade it in and out would be a helpful way of seeing just how close you are.

Now, pardon me while I back track a second and bash this into the ground.

Two tips. Double click for color replacer, and scale 1:1. You know, it's funny how you can miss things that are so incredibly obvious some days.

Anyway, the last time I tried this was still on the OP binedit, and if I'm not mistaken (and I probably am) but either the points were larger than the lines or I had a color match problem or something... because they gave me no end of trouble. Today, however, is a different story entirely.

<center><img src="http://mtm2.com/~forum/images/post_2940a.gif" width="256" align=middle> --> <img src="http://mtm2.com/~forum/images/post_2940b.gif" width="256" align=middle></center>

But this gives rise to two questions.

When you're mapping, do you map right <u>on</u> the lines, or very <u>near</u> them on the outer edges, and do you over spray the outer edge so no 'black' shows up on the model?

Secondly, and this is related, we have two adjoining edges. Thus:

<center><img src="http://mtm2.com/~forum/images/post_2940c.gif" width="600"></center>

Shooting for a perfect seamless join, what is the relation of those two lines to each other? I know it's only one pixel but sometimes that's enough. Do those two lines butt against each other, do they overlap (as in superimpose one over the other), or what? On a solid color vehical this probably wouldn't matter, but some paint jobs can't be taken for granted. To date, on a basic repaint, I'll get them close then jiggle the rear fender mapping to get a better fit. (Man, I always forget the texture shift keys - I have to write that down someplace). Anyway, if we're creating images made to order then that shouldn't be necessary. At least, not ideally so. So, say, I bring those two line drawings together so the outline is of one continuous truck. Now I paint it. When it comes time to slice it back in half does the front get the line, the back, or both? Yes, I'm splitting <strike>hairs</strike> pixels here. But if I'm going to do this, then over-kill is the rule of the day :-)

Thanks again.

> the last time I tried this was still on the OP binedit, and if I'm not mistaken ... the points were larger than the lines

Yeah, in BE.OP the vertices were represented by a 3x3 pixel cross. BE.RB3 uses a single dot scaled to the pixel-size of the texture you're mapping to - much nicer! When I did those falcon textures using OP BinEdit, I had to manually trim the texture created to remove those parts of the crosses that extended outside the area.

Quick note. In either case you could (if you so wish) use the colour replacer to change the vertex marks to a contrasting colour. Then you'll know you have the mesh mapped exactly when its vertices completely obscure all the different-coloured marks:




> When you're mapping, do you map right on the lines, or very near them on the outer edges, and do you over spray the outer edge so no 'black' shows up on the model?

I map right on top of them, and yes, normally you'll want to overspray the edge of your texture shape to avoid the black background bleeding in (usually a one-pixel outline is enough).


> we have two adjoining edges. Shooting for a perfect seamless join, what is the relation of those two lines to each other? ... Do those two lines butt against each other, do they overlap (as in superimpose one over the other), or what?

Good questions. Let me preface by saying I might be wrong but...

Intuitively, I believe they overlap. To divert a bit, with the old BE, it seemed to render the texturedmodel by placing the vertices on the edge intersections between texture pixels, if you can imagine such a space exists, and it always deviated down and to the left from whatever pixel the vertex appeared to be sitting over (so for a square face, a given pixel would be either on the face, or not). I remember this because I had a dern hard time of it with that falcon when I got into the game and noticed my texture mappings seemed to have crept to the left from what the textured model looked in BE, causing drawn panel lines (marking the seams between texture blocks) to expand or disappear. (There was a thread about that somewhere here...)

New BE seems to render the texture on the model with the vertex placed in the centre of the texture pixels, and that seems to match the way the game does it, so it's truly WYSIWYG. So say you have the edge of a face going down a line of pixels, half of each pixel on that line is on the face and half is not. (Yes, now we're splitting pixels in half. Bear with me.) So, that contentious line between your truck cab and bed belongs to both the faces fore and aft of it. That is, if you wanted to join the cab and bed textures together, that line would overlap; if you then chopped it, they should both get it... the faces mapped to those front and back sections share the line between them. If you don't duplicate it front and rear I can almost guarantee you'll get a black line on the edge of one or the other set of faces.

Doh, that above was me. Do not use two browsers to view the forums without logging in with both of them...

When I piece the side texturess together, paint, and then split them up again, I will usually leave a little extra "material" on each section where they join together....Gives me a little ajusting room when I'm mapping the sides.

<a href="http://mtm2.com/~forum/images/post2940e.gif">Scale affects the dot size</a>, by the way. Looks like lines stay at one pixel but dots increase with scale.



That's a good thought too.

> overspray and point position on the pixel

Good stuff. And I do remember the shifting when you were doing it. And yes, there have been times I got the black line too. On those occassions I overshot like Bigdogge describes. But that's been repainting as opposed to working from scratch. I had hoped this paint-to-fit routine could give us more precision, but if you're over-spraying anyway, it can't hurt to leave a few extra pixels on the seam just as a buffer region - sort of the two pixel idea.

> shade to paint

Mal, you'll have to show us how that's done one day. However, wouldn't a gradient fill accomplish the same thing? But hold, shading is a step or two ahead of where I'm going.

Next up, are there any advantages or disadvantages to wrapping around all in one shot like this?

<center><img src="http://mtm2.com/~forum/images/post2940d.gif" width="600"></center>

I recall alp doing this long long ago, and I've heard people talk of wrap around but haven't actually seen or heard anybody detail the process. And by the way, we're getting close to what prompted this latest round of questions. I was speaking to rep fan about a half-way photo texture thing that wasn't resizing at all well, and something he said gave rise to this wrap around idea, which led to paint-to-fit, which led to.... you get the idea.

Phin, I never have tried a full wrap but I always do the sides of my trucks as one piece then divide them up into hood/cab and bed. The only disadvantage to a full wrap I can see is making sure its divided up properly so there are no gaps.

The full wrap is something I should look at, if you've looked at my truck The Ride you can see that the rear green swoop stops suddenly at the end of the bed. That was because a. I did the sides and tailgate separately and b. I wasn't going by a template on the side and misjudged where the swoop would end, its actually supposed to go off the top of the bedside. Now I know better

if you could wrap-map a model in one piece that way, then it could be helpful. I've seen it done to the heads of human models for various rpg games..... the back, the sides of the head and the face were one continous texture.

The main problem I see with it is that a texture like that would have to be reduced in size so much to fit within a 256x256 RAW texture that much detail would be lost.

Of course, you could PAINT the texture that way, but it wouldn't be necessary unless the paint design continued from the sides to the back. In most cases, I just do the tailgate separately.

Welp, I really liked the idea of mapping the textures before painting so that I could look at things on the truck as they progressed and at the same time match the working canvas to the texture canvas point for point. No sooner than I begin, however, and we run into the first snag. Even tho I mapped point to point, the seam where the box meets the door has issues.

<center><img src="http://mtm2.com/~forum/images/post2940h.gif" width="500"></center>

According to our theorizing above, that shouldn't happen. What we expected to get was half of the pixel on the left meeting half the pixel on the right. Instead, we get something like the pixel overlap in a track terrain.

My first thought was my methodology was wrong. I played around from scratch but wasn't happy with the proportions I was getting, so what I ended up doing was using the existing mapping as a guide, then took the pic over a solid color texture. In short, the original shape has for the most part been preserved. But I don't think this is the problem. The points do, after all, line up on the texture.

For the tailgate, the original mapping was terrible and disproportionate so I redid the whole thing and made sure it aligned with the box, then mirrored for the other side, and mirrored the left for the right. There should be no mismatching whatsoever. But... there is.

<center><img src="http://mtm2.com/~forum/images/post2940g.gif" width="500"></center>

Same goes for the hood.

<center><img src="http://mtm2.com/~forum/images/post2940f.gif" width="500"></center>

My next thought was that perhaps the layout button does something to the mapping pattern to give a more accurate shape, but I soon dismissed this because it's the point that really matters. If two spots are mapped to a single vertex, then shouldn't the model pull everything together?

I tried in game with the same results, so it's not a binedit thing.

Looks like fudging the mapping is in order whether you're just repainting or trying for a perfect fit.

On the up side, the wrap around will still be much easier, I would think. Those spots are off, but they're not THAT far off.

==========

For my little wrap image. I actually took the outlines of the truck and strung them along all in one piece. It was something like 1280 wide and 128 tall. In theory, that should slice up and paste onto the mapping texture without a hitch.

And yep, there'd be no point unless the design needed to go from one side, across the tailgate and up the other side. Whether or not it's worth this much thought and bother is another matter entirely... but there are still general considerations too, so no harm yet ;-)

Ah, texmapping... it's an imperfect science. Soooo fun

I haven't played with it that much with BE2 but I remember with the old BE that I seemed to sometimes get different results when I mapped the same face group in what looked to me to be the exact same manner... I was sure I was putting the dots in the exact same locations, but the resultant look on the model would change.

I know that at the raw data level, the texture map coordinates (called U,V coordinates) are stored not as a pixel coordinate on the texture as you might expect, but as a number within a certain range - most commonly real numbers from 0.0 to 1.0 (technically BIN files seem to use an integer number range, but it doesn't make any effective difference), scaled so that (0.0,0.0) is the top left of the texture and (1.0,1.0) is the bottom right. (This is why you can swap textures of different sizes and the mappings stay in the same relative positions - the vertices aren't mapped to pixel coordinates but to relative positions.) I would guess there's some rounding error occurring and BE is tending to favour pixels on one side of the vertex over the other.

You can try mapping it again and see if the result changes (and there'll be subsequent hours of fun if it does ), and/or you could try to force it by deliberately placing the vertices on the right edge of your mesh one pixel to the left of the marks on the texture... but it's probably not worth too much worry.

That's the ticket. I just needed somebody to tell me the incredibly obvious (do things the hard way) and I'd look for an easier route.

<center><img src="http://mtm2.com/~forum/images/post2940i.gif" width="500"></center>

There's no way I was going to fiddle with moving those dots on the texture. So, instead, I used texture shift (ctrl+J / ctrl+shift+j). I don't know if it's perfect, but there's no gap between those two lines.

And in the texture mapper.

<center><img src="http://mtm2.com/~forum/images/post2940j.gif" width="431"></center>

I don't know if it's wise to move those dots now or not. Prob'ly not. It'll just undo what we've just remapped. I'll have to take a moment to think about what's going on here. Before the shift, I had planned to map the back half to a second texture - thinking maybe one texture couldn't share coordinates - but I'm making myself dizzy, lol.

when im going to make a replica, i start simular to D2S. i usually start off by finding a pic that is a semi straght on shot, resize copy paste and then paint over it and keep a copy of the pic for reference. then go and map it to see if it works. if not then i keep touching up the textures until it looks right.