I've not used 3D before; having had only Photoshop Standard, and never really having a need for 3D effects. But I'm experimenting with 3D in the Photoshop CS6 beta, without reading or doing research - just poking around to see what I can do. I'd call this thing a success, since I've been able to create some quite nice effects just by pointing and clicking.
Since I have no past experience with prior implementations of Photoshop 3D whatsoever, I have absolutely nothing to unlearn, and this implementation so far seems fairly straightforward and discoverable. But I do have a few first impression observations and questions:
1. I had to dig a bit initially to find the way to cause the design to render. I found the Render entry in the fly-out menu, and its associated key combination listed there. I also finally found the tiny icon at the bottom of the Properties panel. However, it seems as though there ought to be a "latching" [Render] button somewhere more obvious - maybe up in the Options bar - that you can both push and un-push. Maybe it should change color to show rendering is active. It did discover a way to stop the rendering - I just click on the image - but a more obvious button allowing one to start rendering, showing the current state of rendering, and allowing one to explicitly stop rendering just seems missing.
2. Is there a reason one should wait for the full rendering to occur, if one sees that the result at a given point has gotten "good enough"? For example, in the image above I stopped the rendering when subsequent passes weren't making any obvious difference, even though the status display still showed something like 80% remaining. I had left the image zoomed in on one of the letters and its shadow so as to be able make that judgement carefully.
3. I noticed that the 3D layer returned to what looked like the original "draft" view during design when I subsequently cropped the photo. Is this because I had not allowed it to complete rendering? If so, that's a bad thing, since I was quite happy with the partial result and would prefer not to have to wait any longer. Just this simple design busied my 8 core workstation for a long while!
4. When rendering, I see that the blue grid indicator seems to spend a lot of time on parts of the image in which there are no 3D objects whatsoever. Is this because I've inadvertently set something up wrong (ambient lighting?), or does the rendering process just need to re-evaluate the "nothingness" every time to be really sure there's nothing needed there? It seems like the rendering process could be VASTLY sped up if it just could come up with a bounding box early on and eliminate areas where there is simply no 3D info to be rendered.
Nicely done for a first try.
Waiting for the rendering progress to finish is purely a subjective call based on the level of complexity in your 3D layer and the output destination. In the 3D preferences, you can change the 'quality' of the Ray Trace results. Setting the value lower results in less passes so you'll get a 'dirtier' result but it will be consistent. When you stop the process before it's finished, you might have areas (seen as tiled areas) that more fully renedered than other areas.
Many interactions, including crop, will change the displayed 3D layer back to GPU interaction rendering. It doesn't matter if the rendering went ot completion or not. We haven't implemented a way to store and view multiple renderings (except via the History panel) automatically. You can do this manually by copying layers, but it's not ideal.
One thing to speed up test pass renderings is to marquee a small selection and then render. Only the selected area will render. The reason you're seeing render progress squares away from the 3D meshes is that the whole 'scene' is getting some amount of light bouncing into that area.
Thanks, SG. That was very insightful.
In the interim I've played a bit more and noticed that if you make the 3D layer into a Smart Object, the rendering can be kept rendered through subsequent operations, and it affords some ability to move rendered things around more freely and alter them (e.g., via smart filters). Only problem then is that it's a longer jump from seeing the 3D object in the context of the image it's being rendered in front of.
It brought to mind another question, unrelated to 3D: Once you've made something into a Smart Object, is there a way to undo that (outside of going back in the History)? I have never looked into doing that before.
Unfortunately, there's no command to 'unwrap' a Smart Object (SO) layer back to it's original layer type in the parent document. What you have to do is open the SO layer, and then duplicate the source layer(s) back into the parent document.
Europe, Middle East and Africa