I have a logo which I have built and optimized to run on a realtime engine. The artists working with said engine need a single texture of baked AO that maps correctly across the separate elements as they exist in the engine.
What I did to get a combined UV map which allows for the single texture was to select all of the objects (in an iterative save) and do a combine + delete. The resulting UV map was exactly what the guys needed.
HOWEVER - the fbx they bring in has each element broken down into 3 pieces (faces, bevels, and sides, so they have more control and can texture the elements separately. When they attempt to bring in the single bakes texture, and apply it to one piece of the logo, it puts the whole texture there rather than just the section containing the UV’s for that part of the object.
Obviously, this is happening because the separate elements have different UV’s than the combined object.
So, how can I make it so that this works as I described above? One of the viz artists insists it is an easy feat in Maya (which I have a hard time believing if only because NOTHING is every easy in Maya), so there must be a way to do it in Cinema, no?
I don’t have any informations other than you have provided here about the project, so you might have left something out in the initial post—I guess.
I have to admit that you lost me in your description of the workflow. You merge everything and the FBX they use has all of the sudden everything separated again? Perhaps they use a previous (before baking) file and you have merged it, but for what use if they work with a separated one. I just try to understand, not to critique here!
I don’t get the point why someone would separate everything first (which might be fine)—only to use then a second layer (but merged) for the AO? In other words, why not stay separated with the AO as well and deliver three (but smaller) AO textures. With the same final resolution, the textures won’t be that much bigger, what you save is a complete second UV set of informations in that way. Realtime is optimizing data, right?
As I said, something is not logical in that workflow.
I do not discuss other apps here, especially not to make less of others. Taking their word for it, if that is so easy and they have an FBX, which is kind of a house/native Autodesk file, why they don’t just take this (Maya) as nice authoring filter for their workflow.
Again, please elaborate why a merged file separate it self magically to caps, bevel and sides, and I might understand the first step—then we might be able to go further.
Well, this is for me a new job, and for the company I work for, something of a new position. Let me try to explain the bigger picture briefly.
The company, Reality Check Systems, produces most of the flashy 3D graphics packages for sporting events the world over. The software used to drive these packages is a realtime 3D engine called VizRT. It is essentially a GPU based realtime renderer, not at all unlike a game engine. Many of the elements the Viz Artists are using were created first in other packages - most often C4D though there are Maya files occasionally as well. One of the tasks assigned to me is to take these Cinema files - sometime created in house, and sometimes provided by the client - and ‘optimize’ them to run on Viz. This means mostly bringing the polygon count down as far as possible, eliminating any intersecting geometry, deleting any back faces, and setting up the hierarchy to make it easy and organized for the Viz artists.
So, let’s say I have a simple logo with 3 text elements and a symbol which sits behind the text. Once I have done the poly reduction, etc, I then need to set up the hierarchy. What works best for those guys is if each extruded object has the sides, bevels, and faces each as their own object - pretty much exactly as it happens when you make an extrude editable in Cinema (though I make all 3 same level children of a single null). This way, the artists can apply their textures to elements separately, allowing them to do the various cheats and fakery they do to get end result you see on TV.
Once I have sorted all of the geometry and hierarchy stuff out, my next step is in baking out any textures used. So far, I haven’t had any textures to deal with, just simple materials which can be easily created from within Viz. But, the one thing they always do need is an Ambient Occlusion pass. The way they want it is as a single 2048 x 2048 file of the entire logo, so that all parts of the 3 text elements and the symbol are represented. They then take the texture as a whole, apply it to each of the separate objects in Viz, and map it as ‘Vertex’.
Now, the only way I have been able to get a single texture with the UV’s for all objects present is to select all of the geometry and do a ‘connect objects’ command. This allows me to get the texture map the way it needs to be, BUT, it does not map correctly to the exported FBX file I have sent them with the appropriately separate bits. So it seems that what I need to do is find someway for the separate objects to have the values of the combined object’s UV map. One of the guys did a quick test with Maya, and it worked just fine. However, part of my reason for being here is to help the Viz guys streamline their workflow, and not have to use time for these tasks. I do not have Maya, and don’t know how to use it, so pulling that in for this step… I suppose that could happen, but it would be much better if I could keep the workflow on my end all in one program whenever possible. BodyPaint and UV mapping is an area of Cinema that is new to me, despite my 13 years with the program. so I feel there is just a simple procedure I am not aware of.
As for why FBX, Viz needs either that or an OBJ for 3D geometry… though the next point release of Viz will support .c4d files natively, and that’s kind of exciting.
The idea how the “real-Time” engine works is important to know. My main question would be, if you use one texture several times—will it be loaded into the GPU one time or each time (multiple)?
If only once for the scene, I would go with a shared texture and set up the UV map accordingly. As you mentioned you experience with UV Edit, please check out the chapter Wizard>Single Material Mode ... for several objects.
The next question would be—if their engine supports multiple UV maps per single object. If yes, you can deliver anything you can do in C4D. If not, you must go with the idea above of split.
If they have single textures for each object, but shared textures for AO, then they must have an option for multiple UV map information.
I have attached a file where I use one texture, which can hold the information for all four objects, separately. (Following your four part description cap, bevel, sides, background). Please note that in the BodyPaint 3D>UV Edit>Layer>Fill Polygons (or outline P.)—you have a option to mark the areas where you have moved the UV polygons on this specific texture. This helps to manually organize a texture space for many objects.
Name the UV Tags accordingly to your objects to stay on top of the game, not naming them is OK, but for your sanity, do it.
Thanks Dr Sassi. I got momentarily pulled away from the logo work to optimize the geometry of a robot arm, but I will refer back to this and study your answers in depth when I circle back around to the logo next week. Thank you very much as always for sharing your immense knowledge of this program with us mere mortals
Thanks a lot for this nice feedback, Will.
Knowledge is nothing if I can’t apply it. So, let’s find next week what would be the best possible way to do it. The more I know about a problem, the more I can exclude stuff that might not work.
As you know me, I like to start from the origin of a problem, not solving something half the way, which is most likely never as good as starting from scratch. In this case, the real-time sets technical preferences, and places artists comfort on second place. Which makes it perhaps easier to decide. The product counts here, not the way predominately.
Ok, so I am about to get back into this. To answer your question above, a single texture can be applied to numerous objects in a scene while only being loaded onto the GPU once. This is why they try to get single textures whenever possible. Normally what they do is take an AO bake and a reflection map (which they pull from a large library of files that were made in Photoshop) and load those into the scene, applying to objects as necessary and combining them with the extremely simple (3 channel) materials that one can create in Vis.
Right - off to have a look at the file, and put the theory to practice - wish me luck!
hmmm… Ok, so here’s what I have done. I initiated the Wizard, and made sure that Single Object Mode was enabled. The result was what I was hoping for, clicking on each object’s UV tag showed me its UV Mesh in a different section of the texture view. Now, I needed to get all of these individual maps combined into a a single one for baking. I selected all of my objects and performed a ‘connect objects’ command. Clicking the UV tag there showed me the combines map of all the objects, and it looked exactly I wanted it to. So, I did a quick ‘Bake Object’, but the resulting material is still not mapping correctly to the individual pieces, despite those pieces having now the UV’s as described. Instead the entire material maps over each section rather than the material respecting the UV’s.
I can’t share the project I am working on for NDA purposes, but I have put to gather a mock up to illustrate the results I am getting so you can perhaps tell me what step I am missing.
... To answer your question above, a single texture can be applied to numerous objects in a scene while only being loaded onto the GPU once. This is why they try to get single textures whenever possible.
Hi Will,
I get that a single texture for a single multipart object is the target. The “Wizard” allows for such set up in the simplest way. Besides that, as my example above shows, it needs some work.
If I get you current challenge correctly, you clean up models and reduce the density of those as much as possible. In other words, you have a new model, which needs certainly an inspection of the UV map anyway.
I would check each part with a color checker board, if the surface-pixel density is sufficient, then take the suggested image/texture size and sort the UV to a proper area of it. After using the “BodyPaint 3D>UV Edit>Layer>Fill Polygons (or outline P.)” to mark the area in the texture, you can place the textures to the parts. Perhaps get familiar with the process of projecting a texture (e.g., sphere or frontal, etc) and then use the Assign UVW Coordinates to tag.
One question, Will, why do you connect the objects? I think this is a step that makes no sense here. Is the “Single Texture” option in the “Bake Object” and (!) “Keep UVs” not working for you? Select the objects all at once, the ones that that have to share the same texture.
The file that you have attached above, shows a different UV map, or do I miss something?
If you use the Replace objects, you get separated Object, but with the single texture on, a single texture after all.
My understanding so far is, that you have to provide the model-mesh in separated parts, but the texture files in one single texture for all parts of the complete model.
My best wishes. Let me know if that works for you.
I see. My presumption was that since I couldn’t see all of the UV’s together as a single map, (it only lets you see them one at a time) the connection was necessary to combine the pieces into one. I did replicate you settings in that dialogue box, and did, at last get the desired result.
Now that the the test has been successful, lets see how it scales up to a full scene.
Thanks again so much for your help on this, I’ll report back if I have any further issues.
Great and thanks for the information, Will.
Information is key, so I ask.
I think we are a good step forward, crossing the fingers that the workflow will scale nicely to the larger scenes.
