How can i combine Normal and Bump maps

[stuh505]

You have to put a Normal Bump wrapper in the Bump slot anyway, and there is a map slot for both a normal map and an additional bump map there..

[Ninjas]

use the Nvidia PS plugin on the bump (greyscale), and then screen the layers together. There can be some weird artifacting (inside out normals, etc).

Just checked out you website. Some of your renders make the models hard to see (the GI renders). VERY tight work. There is a good Polycount thread on this very topic.

[rv-el]

Thanks guys. I will check into the Nvidia Plugin. Does it reference the model at all? seems odd that that would work. I wish there was a way to combined them in max's render to texture. something kinda.... auto-magic :)

I'll try to find that polycount forum post too. Hard to find it.

The reason my stuff looks so dark is because i use a bright monitor when i make my art. Its a damned Sony Trinitron.. its set at 30/100 for brightness. even at 1/100 its STILL super bright. I should put a warning on my web site for people to crank up their brightness. This has been driving me nutts forever. I can't wait to get a new computer with some new LCD's and try to figure out some sort of standard. To me it all looks really good.. on other monitors some peoples stuff looks super dark, some look fine.. so i basically stick with my setup.

[EricChadwick]

Courtesy Pior Oberson (http://www.pioroberson.com/extras.php)...

http://www.pioroberson.com/images/minitut_screwyou.jpg


And here's the method I use to set the gamma for our monitors. Makes a huge difference.
http://www.aim-dtp.net/aim/evaluation/gamma_space/index.htm

[rv-el]

I dont know that this is too good for a complex character model. I have uves turned sideways and rotated around at odd angles to maximize my UV space. The results from this plugin seeing as it references no UV/3d space can only be... well.. "straight up and down".


I need something that will render out a combination of the 2 maps.

I'm not quite sure how to work Ati's normal mapper. so far the program looks like it was created only for the people that made it :).. not too friendly.

[rv-el]

Thanks for the gamma test thing. I am doing it right now and on the black point test i'm finding out that even with Contrast at 100% and Brightness at 0%, the black on the desktop background is not even close to the black on the borders off of the scan. At 0% brightness the white looks ok.. but its hard for me to tell if it should be brighter.

I don't know what to think.. i'm pretty angry about this concidering i've been doing all my work on this monitor for YEARS and when i look at it on others computers, most of them look fine, but on some of them its dark.


edit: I just found this in about my monitor problem, and how to fix it (painstakingly opening the thing up and solderin gon some new transister of sorts)

----
Many Sony Trinitron-based monitors manufactured around the 1999-2000 time frame develop a brightness problem over time which substantially degrades the picture quality. Some models which exhibit this problem are the Sony G500 and the Dell P1110. The following "fix" has helped me and many others get the brightness and color back to near factory specs. The main signs that you have this problem will be the absence of a good black level - even with the brightness set to 0, the monitor will show blacks as bright greys. [Disclaimer: this "fix" is not really the correct way to fix the problem, but the results should be quite satisfactory
----

Excellent just excellent.. soon as i get a job, soon as i get some pay checks.. dual LCD's here i come.

[Ninjas]

Have you tried putting the bump into the displacement slot when you do your normal map? I have been meaning to test this. I have no idea if it will actually work.

[rv-el]

A piece of information that may be imperative here is that i no longer really have the hi resolution mesh. I broke my character up into 3 pieces so that max could handle the hi poly counts on the OBJ's coming from Zbrush. Then i composited this map in photoshop, then attached the model back together, and applied the compiled map.

If i make a bump map and apply that with the normal it renders perfectly.. now i just need it to use the same thing it does to create that image, and reverse itself into crapping out a normal map that reflects that final result.

may be this isn't possible.. if i have to have the hi res geometry no matter what then i'd have to go back and re break up the model, generate, recompisite in PS again. :(

[EricChadwick]

It's really simple, we do this all the time, works great.

1. Put the grayscale bump map on your low-res model, then use RTT to bake a new normal map. These normals will be relative to your UVs.

2. Take this into PS, fill all the empty black areas with flat blue (127,127,255).

3. Put the geometry-derived normal map in a layer below the height-map one.

4. Set the upper layer to Overlay.

5. On the upper layer, choose the Blue channel and do a Layers adjustment on it... in the Output Levels change the 2nd one from 255 to 127.

6. Merge Layers and run the NVIDIA filter, set to Normalize.

Done!

[rv-el]

Ok, how do i get that out of RTT? Do i copy the low poly model, then apply the bump map, then use RTT on the original low poly model and reference the copy and render?

After i have this, it should be a normal map, so why do i need to use an Nvidia PS plugin after overlaying?

Thanks guys, its a big help

[EricChadwick]

Ok, how do i get that out of RTT? Do i copy the low poly model, then apply the bump map, then use RTT on the original low poly model and reference the copy and render?
You want to combine a grayscale bump map with a highpoly-extracted normal map right?
One way would be to UV the highres guy, put the bump map on him, then use the lowres guy to extract both geom and bump at the same time. Simpler way though is to not UV the high, but extract just the geometry normal map, then use RTT again on the lowpoly, with just the bump map on himself and output that to another normal map, then combine the two in PS.

After i have this, it should be a normal map, so why do i need to use an Nvidia PS plugin after overlaying?Because if you don't renormalize, your normals in the combined map will not be a uniform length of 1.0, which can cause problems with the specular.

Good luck!

[rv-el]

Thanks for the help! here is what i did layed out from begining to end.

1. Copy your low poly model twice. Dedicate one as the Working Model and one as the Bump Model. (probably works with the original model for working model)

2. Make a new standard material and place your bump map in the bump slot. Apply this map to your Bump Model.

3. Select Working Model and open Render to Texutre (0 for shortcut). Assign Normal Map, Make sure your using existing UV's and pick the right set. Click the [Pick] Button and select your Bump Model. Reset your cage and set it up however.

4. Click render and save out the map. It should be a clean normal map version of your bump.

5. Take this into PS, fill all the empty black areas with flat blue (127,127,255). Make sure it is lined up over the top of the normal map you want to mesh it with.

6. Set the upper layer to Overlay.

7. On the upper layer, choose the Blue channel and do a Layers adjustment on it... in the Output Levels change the 2nd one from 255 to 127.

8. Merge layers and then run the Nvidia plugin on this layer.. Make sure you click the Normalize only radial button in the Alternate Conversions area.

Note: You can shortcut the second half of this process by applying a standard material to your Working Model that contains a Normal Bump in the bump slot with both your original normal map and your bump applied. The problem i ran into is that this works but the normal map came out with a lot of artifacts/errors in it.

[EricChadwick]

this works but the normal map came out with a lot of artifacts/errors in it. Don't pick the Bump model as your target. Instead, select the Bump model, invoke RTT, and don't use "Pick" (because the Bump model is baking itself). This avoids any overlap/misses problems with a Projection cage.

Basically, you don't need to copy the model. You only need Projection if you're baking from one model to another (high to low).

Does that make sense?

[Brain Trepaning]

Makes sense to me and very good info, thanks a lot!

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.