TR4B White Line Fever

[slider38]

Hi Guys, Perhaps someone out there can figure out what is causing a white line to appear on this engine, a TR4B that Snoopy was kind enough to work up for me. I have been trying for a couple of months to figure out what is going on. In game mode a white line appears where the old engine section meets the new section.

TR4B1.jpg

But as you get closer to the engine, the line begins to fade. Get close enough, it's gone altogether.

TR4B3.jpg

I've tried everything I can think of to make this white line go away. It does not show up in NifSkope. I can't fix it as I can't figure out what is causing it. I am hoping maybe someone out there has a new idea as to what is going on here. I've simply run out of things to try. Thanks.

[NSrailfan]

I suspect this to just be a minor design flaw. In NifSkope it looks okay, but we know this game loves to mess with us. My solution would be to pull it into Blender and combine the vertices where the cab used to be. The long hood and short hood being to separate pieces rather than just one long shell on the body may be where the problem lies. This is just a guess though, but would be worth a try.

[laloha]

Well I am sure not a reliable source, but I might go into NifScope and possibly move vertices along that line towards the other body part. Just a fraction. Maybe there are similar vertices on the opposite side. If so check if there is a slight difference, provided the other side does not have the white line too.

Maybe you already did this, then NSrailfan sounds like an excellent choice since you have that ability too.

[snoopy55]

In speaking with Barney I tried a couple of colors without adding shadows and with adding shadows. (and I thought my rotten day was getting better....)

Attached are your DDS and two others, one using a green I cooked up and one using a blue pulled from your DDS. The green is 0,91,0 and the blue is 33,36,57. All I can see in NifSkope on those two is a few pixels flickering when the engine is tilted. With your DDS I see the white lines.

Now, major question...... How do you layer your DDSs? Yours is RGB, 9 Layers while mine are RGB, 11 Layers. Since I do not have an original of yours, which I hope you kept, unlayered, I cannot correctly check it out, but I did a copy visible and pasted it onto a new alphad image and then layered it BC2 / DXT3, with Generate mipmaps checked when I saved it. This is how I have been doing them since I started this and how my tutorials have been written.

Post the three colors you used and let me play with them. I notice a lot of color bleeding in your DDS.

I'm sorry if I sound hard in these posts, the last few weeks have been hard on me both physically and mentally, and the work in this game is the closest thing I have to a release.

[slider38]

Snoopy, I will test the diff files that you sent me. As to the colors that I used, if you look at the TR4B_diff you will find that there are a lot of colors used. This engine had no shading other than the various shades of light gray in the original diff that you sent me. Consequently, as I colored the diff, I replaced each shade of gray with a slightly different shade of blue. The main colors are Blue - 38, 36, 63 Gray - 137, 142, 156. Again you will see variations in the gray.
As to my diff.dds files. They are created as tga files with my Paintshop Pro paint program and converted to a dds format with DxtBmp, a converter I've used for all train work. When you go to save a file in the converter, you have a number of file formats to choose from:

DxtBmp_file_formats.jpg

I have seen some of your files occasionally posted in what the converter says is DDS 888-8, a 32-bit format. Those files were huge. I do know that my files came out in 9 layers as I would see that when I went to Gimp to create a normal. I really don't want to use anything bigger than DXT 3. My files have a way of getting kind of big already. I did notice the bleeding in this diff. I don't know that I have seen much of that before. I'm not sure if that's a function of the colors I used or what. I will test the diffs that you sent me and see how they work.

NSRailFan, I appreciate your suggestion. I have gone through NifSkope and changed the vertices where the two sections meet so they would be exactly the same. It made no difference. Laloha, I have removed the rear section and reattached further up the skin - no difference. I have tried to cover the joint with a small plane in Blender. I have tried covering it with a narrow decal. I have spent untold hours, fiddleing, and finagleing, testing and re-testing. This is just a super pain in the butt. Sorry, but I am pretty-well fed up with it. Oh well, I'll try some tests with it tomorrow and see what these diffs do. Thanks guys.

[slider38]

Bobby, you are the man. No more white line. I replaced the white background on the diff with a dark gray, at your suggestion, and the white line is gone. I did it for the A and B units and they look great in game view now. Thank you! Thank you! I have a bit of work to do getting the files reorganized again, but that's nothing. Thanks to all who tried to help me with this problem. I really appreciate everyone here for being so giving with their time. The world's greatest website. Now, back to work. I'm testing a new power plant and I need to get these TR4 files cleaned up.

[bobby]

:yippey

Bobby, you are the man. No more white line. I replaced the white background on the diff with a dark gray, at your suggestion, and the white line is gone. I did it for the A and B units and they look great in game view now. Thank you! Thank you! I have a bit of work to do getting the files reorganized again, but that's nothing. Thanks to all who tried to help me with this problem. I really appreciate everyone here for being so giving with their time. The world's greatest website. Now, back to work. I'm testing a new power plant and I need to get these TR4 files cleaned up.

Your welcome . I know this problem was around for ages and somewhere on the forum it was discussed and resolved. Not sure but i think it was when snoopy 1st created Construction kits.

[slider38]

Ok. I've finally gotten rid of the color bleeding. I think you were spot on Snoopy when you questioned the number of layers being used in these diff files. I cut the diff down to 512 and tried saving the files as dds files in DXT3 format. I still had color bleeding. I saved the in DDS 888-8 32 bit format and the bleeding was gone. Having reduced the files to 512 the resulting diff file is the same size as the original 1024 files I started with. That's not too bad. Thanks again for all the help. Here is the result:

No_color_bleeding.jpg

[snoopy55]

You might want to try doing a search for psp 9 dds plugin . Barney and I researched that when we first started playing with the games graphics.

[slider38]

That's a good suggestion Snoopy, but I own PSP 5. I don't think it's been supported for the last 10 or 12 years.

[laloha]

DXT5 and DXT1 were tho two processes we used and is still used for the car racing offerings I was involved in. Can't explain the reason why not using DXT3 instead however. The DXT5 was used for all the cars plus some other files, where DXT1 for the windows, helmets and other files too. Tried the DXT5 here and works fine from what I can see. Might try this to see if there is any difference when problems crop up.

[slider38]

Actually I tried doing this diff in DXT5 format, but I was still seeing some color bleeding. I have worked with 1024 x 1024 files before and never had a major problem. I think that the colors involved in these engines seems to accentuate the issue. There's so much contrast between the very dark blue and the gray and gold. My guess is that I had some bleeding in some of the other stuff that I have worked with, but it just never showed very much so it didn't really stand out.

[snoopy55]

Some things can really peeve a person off!!

You had this white line problem...Bobby had seen something of it but could not remember where... I'm going thru the CKs because of a minor XML problem..... And what do you think I came across.... viewtopic.php?p=5698#p5698