The meshes that get generated on caps, bevels, and extrusions are adaptive, but there's some things that can make them smoother.
From your capture, I'm assuming you don't want this as the final cropped output, and you're just zooming a screen capture to show the rendered detail. What else do you have as part of that 3D layer extrusion? Is the gray surface an extruded circle from selection?
Is there a way for you to point me at your source document? You can send me a private message if you don't want to distribute publicly. Or create a quick dummy comp that still shows the problem.
I created by useing the eliptical Selection tool, Fill then extrude from selected layer. What would be the better way to creat objects that reduce this jagged edge and surface?
I did the same to a 600px Doc, 3000px and 6000px... seems the higher the Px count the tougher psd has modeling?
Okay, so you extruded from pixels... I'm not saying that's wrong, but does it do anything different if you extrude from a shape layer?
It occurs to me that to make the 3D mesh Photoshop has to trace the edges of your pixels in the layer you've provided, using line segments. There may be limits to how smoothly it will do that. I don't know whether there are settings that can affect this.
On a big image I'm able to reproduce what you're seeing when I extrude a pixel layer. I find a shape layer seems to do it as well, but differently - more uniformly.
Seems to be good for making cupcakes or maybe Reeses Peanut Butter Cups.
Ok. I tryied doing it from fro elipse tool....Little better but still has ridges...I see yours is the same. Is that due to px size?
Or is it best to import smart object from ai?
What would you say is the best practice here? The rendering you made last night look smooth. The one with purple caps.
How did you create that?
That was just a much smaller image (just 800 x 600 pixels or something like that). As you have seen I'm also seeing more choppiness if using very large images.
I'm not a 3D expert, I'm just trying to learn the limitations here as well. But at least I've been able to confirm it's not a problem with your particular system.
Your best bet will be to continue to engage Steve, the Adobe expert. He ought to be able to follow in our footsteps here without your having to send him a file.
I suspect were right here to assume the more pixels the more PSD has problems working the math. So maybe to get somthing high res say above 1000-2000px is a limitation of the software.
Hopfully Adobe will Chime in?
Hi Steve and Noel,
When Ps generates meshes via extrusion is needs a closed path. So if you have a pixel selection, it is converting that selection to a work path first, (using the Make Work Path tolerance) then generating polygons. If you want better control over the amount of points, and thus the extrusion face topology and smoothness, you really want to be explicitly working with path data before extrusion to better predict the results.
The Work Path tolerance also explains why you get more lines with a higher resolution document. The work path gets more points with a higher image pixel count but same tolerance value.
For the simple cylinder case that you have, a constrainted ellipse tool shape layer (not a marquee selection) should give you 4 points on the path and a smooth resulting extrusion.
For shapes that are more complex (say text based extrusions), image resolution (and font size) will also come into play. The higher the resolution for these the more polygons get generated resulting in less faceting along a 3D shape's profile edge.
Image resolution for 3D layers is generally good around 1-2K @ 72ppi (assuming a fairly capable GPU, 1GB VRAM). A few more factors do come into play, such as texture map size, document bit depth, CPU cores, and available RAM.
I'll see if I can post a few images that show the differences regarding the extrusion surface with selection vs. shape layer sources.
Here's a quick breakdown for circle extrusions. I've overlaid the path used for the extrusion on top of the rendered results. These were all done at 512px X 512px.
This image is the constrained ellipse marquee selection with Work Path tolerance at .5 case
This image is a constrained ellipse marquee selection with Work Path tolerance at 1 case
This image is the constrained ellipse shape layer case
I'm a bit confused by your response.
Daniel Presedo wrote:
1. A Path will always yield best results when generating 3D
2. Document size is not really a factor, having said that here is an image I made to show you a best practice for these kiinds of mesh creations.
Concering 1., what are you trying to say? As I show above, 3 paths are used resulting in differing outcomes. The method of generating the path are different, but they are all paths.
And 2., if document size is not a factor why do we see a differences in the rendering depending on the base document size in Steve's, Noel's, and your example? How does one put a 2000px 'R' into a 500px sized document? What if I want to use reeditable text and don't want a pixel based letter; can I use any document size for my source and get the same mesh quality?
I will try to clarify, and I don't want this to degenerate into too much detail that will confuse everyone, but yes you can make a terrible PATH and have terrible results.
1. Paths make better 3D. Yes, the quality of your Path matters. As a rule a PATH is better than PIXELS.
2. 3D is not bound by pixels, so document size is irrelevant when it comes to creating/rendering 3D. Once you have the PATH and create the model you desire, document size is irrelevant.
That example above is a workaround to generating higher quality extruded text. This works with fonts that do not have a high quality path ( most fonts you purchase will be of high quality.) However, for some free fonts we have to generate a Mask outline and that results in lower quality > this workaround above helps in that case.I do not think this workaround will work with SHAPES or PATHS, you need to still do what you outlined above and make a better quality path
Just so it's clear, I created a new large image (e.g., 6000 x 4000 pixels), used the Ellipse Tool constrained with the Shift key to create a circular path, then clicked the Shape button to make it a solid shape layer. Then I extruded it using New 3D Extrusion from Selected Layer. I added a bit of inflation to the cap simply to make it more cupcake-like.
When done in an image with a fairly large pixel count, the extrusion has the vague appearance of a cupcake (sorry I posted such a dark image, but look closely). The base has "ridges" or "folds" showing.
Are you saying the path quality being created by the Photoshop Ellipse Tool is poor for use in a 3D shape?
Here's the PSD document if you want to examine it more closely.
Thanks for providing more info.
Daniel Presedo wrote:
2. ...Once you have the PATH and create the model you desire, document size is irrelevant.
So what you're missing in this thread and question is, "before creating a model, is document size relevant?". We have a couple clear and simple cases that document size, before extrusion, is relevant.
Noel, I'm asking the developers about the scalloping on the circular path for the higher pixel count case. Hopefully I'll have some feedback to provide next week.
It's kind of cool for making slick new recycle bins, though.
One thing I learned in this exercise is that once the extrusion is done, it doesn't matter if the image is made very small, the folds / scalloping remain.
I wonder if someone got a sign backwards, and the scalloping is just backwards from the curvature that's supposed to be there.