I have a hard time getting blurry text in the background with a camera and depth of field look "right".
Because I need to use latex math equations, I can't use built-in text, I need to either use AI files, PDF files or a hi-res version of the text exported in png from photoshop. The results are exactly the same though.
In the example here below, I used:
- 1940*1080 composition
- A camera (-1500 back in the z axis, Zoom 1500, DOF on, Focus distance 1500, Aperture 800, Blur level 100%)
- The first text is at 0 on the z axis
- The second text is at 250 on the z axis
- The third text is at 500 on the z axis
- The fourth text is at 750 on the z axis
- The fifth text is at 1000 on the z axis
- All text instances have originally the same width (3840px) and are scaled just the same (46%).
- All text layers are in 3D.
- The image is rendered with the optimal settings.
But the text layers seem very "pixelated" after I push them forward over 250 in the z axis. It doesn't seem natural at all passed the second layer. The plus sign is the most obvious in terms of degradation. The results seem more fuzzy than blurry.
What am I doing wrong?
If nothing in particular and it is supposed to be that way, what can I do so that the blur/bokeh seems smooth at all distances?
As it is, it doesn't seem THAT bad because I gradually pushed those texts backward and the sense of depth of field is there alright. But if I have, let's say, both the first and last text alone, it doesn't look like one is behind the other, only that one is fuzzier than the other. That's a problem since I have to move the camera rather often in my animation. The sense of depth should be there all the time.
Any workaround that would enhance the result (more blurry, less fuzzy) would be welcome.
Here is an example of something that is approaching what I'm looking for:
The bokeh effect induced by the depth of field seems to smooth and "sphere" things out rather than "square" them.
Thank you for your help and input.