OK. It sounds like you’ve already got at least a partial solution going, that’s good.
As for the image, it is fairly large. It's like 14" wide at 600 dpi. About 140 MB file.
By @coolmodernist
If that’s the size at which you need to print it, and that’s also what the dimensions are in Image > Image Size, then it sounds like the current resolution should be enough for that print size. But the examples posted earlier look much lower resolution than that, so what were the pixel dimensions (height and width in pixels) of the image as it came out of the 3D application, before any resampling in Photoshop?
A quick question. What od you mean by upscaling? What kind of command and process can I try out? Please let me know.
By @coolmodernist
Upscaling is the same as “upsampling” where you’re increasing the number of pixels in an image. Fortunately, you’ve already started down that road because you used a Resample option (Preserve Details) in Image Size. For a challenging image, some would test Preserve Details and the Bicubic options to see which one looks the best, because the best solution is not always the same one depending on the image and whether you’re increasing or decreasing the number of pixels.
For many years, the Resample options in Image Size were the main ways to adjust the resolution of an image. Recently, Adobe has used AI to add more ways to increase resolution. They can sometimes be better than Image Size/Resample, and like the Resample options, each can have a different effect on fine details:
Super Zoom. Choose Filter > Neural Filters, enable Super Zoom, set Output menu to New Document.
Super Resolution. Super Resolution is available only in Adobe Camera Raw or Lightroom, so it can be used only on file formats compatible with that software. It is not available for the Camera Raw Filter applied to a layer in Photoshop.
If you’ve heard of non-Adobe upsampling/upscaling apps such as Topaz Gigapixel AI, that’s a similar thing: Attempting to improve how increased resolution looks by applying AI/machine learning. These new AI methods also need to be tested with line art/synthetic art to see which one works best, because I’m not sure how much each AI engine is trained to properly handle details in images that didn’t come out of a camera. They are typically marketed with examples of photographic real world scenes.