Depth of Field: KS vs. PS

[cjwidd]

I'm curious if there is a general preference (consensus) among the community regarding DoF in Keyshot or in Photoshop. Keyshot has a nice tutorial about using the depth pass to do DoF, which seems to encourage the latter. Obviously you should do what is right for your workflow and, in many cases, what is most expedient, but nevertheless.

[TGS808]

I do it in KS 100% of the time. For me, there's no reason not to and the results always look better.

[cjwidd]

Right, my initial reaction was to do DoF completely in Keyshot, as I prefer to stay in one software whenever I can. However, not having to re-render, and the non-destructive nature of DoF in Photoshop is very compelling.

[soren]

Just to add a technical point here, the two methods of adding depth of field are quite different.

Photoshop does it using a post-process. This means it does not have access to the full scene when doing so and it will be less accurate than what can be achieved in KeyShot. Curved reflective surfaces will enhance og reduce the depth of field effect of the reflections depending on the curvature for example - this will not happen in Photoshop. Also, if you focus on something behind a mirror an object in front at the same relative distance will also be in focus in KeyShot (correct), but not in Photoshop.

The results in photoshop are significantly faster to get without noise though, since it is a much simpler computation.

Søren

[cjwidd]

Very good, thank you for including this information!

EDIT: Just as a workflow note (in case others may benefit), in can be helpful to do DoF adjustment in performance mode before entering Product or Interior rendering mode for the final render - reduces much of the headache of locating your object(s) in a cloud of blur, etc.

[mattjgerard]

Also in my small world of rendering product images with transparency, the depth maps that can be used in photoshop don't do well with high contrast edges, and overlapping objects. Since PS doesn't know what is hidden by another object, it can't extrapolate what that blurred part should be, and gives odd sharp/blurred edges that need to be manually fixed. I am all about flexibility and love the idea of doing the DOF in post, as my "decision by committee" approval chain quite often comes back with multiple opinions on DOF and such things.

That being said, I have found that render layers helps much, if I really need to be able to tweak the DOF in post, especially if it was a really gnarly image to render (lots of practical lights, cloudy plastics, LED's behind cloudy plastics, etc) that I don't want to re-render every time someone changes their mind. The clown layer pass has saved me hundreds of hours of re-rendering to change colors and materials, and now I am figuring out how to use render layers and even queue up multiple renders with sets to expose different objects in the background by turning foreground objects off, then rebuild the scene in PS and have full control and exposure of each object for the DOF to play with. But that's not perfect either as you give up the interaction of light and reflections for those objects.

No perfect answer, and each image is different.

[cjwidd]

These kind of tips are the reason I come here, thank you so much for this^