Just how much worse is the image quality if I don't remove the olpf for vr180?

[Edith Blazek]

I'm finding the raptor x ticks all the boxes for what i want to do in regards to resolution/framerate/size, but I heard it's really soft for vr180 with the dual fisheye if you don't remove the olpf. I heard some people say it's softer than an r5c with but i also people say it's still softer so I'm thinking why bother? Does it really improve it that much removing it? How is it on apple vision pro with the olpf (I don't have one with me so that's why I'm asking)? And before you point me to the Ursa cine immersive, 1, it's big, 2, it's not particularly sensitive to light, 3, it's close focus is terrible and for what i do i require intimacy, and 4, no fiber live streaming, so no dice.

 

[Phil Holland]

RED's general strategy isn't a "ruthlessly sharp image". And as you mention, the OLPF as well as minimal signal processing are parts of the reason why. This provides decent protection from moire and aliasing while providing a somewhat film-like image draw. And the good thing is you can indeed sharpen the material in post if you desire.

If you do find a camera without a OLPF, and they do exist, you will get slightly more detail, but inherit some of those potential artifacts.

Generally for work I've done with that lens and dual camera setups, we do a bit of post processing to crisp things up as desired.

R5C is a more processed image from the get go due to the DSP and image pipeline. I also own this camera.

 

[Alex Zeder]

... the raptor x ticks all the boxes ...

This is more a hypothetical consideration because the V-Raptor and DSMC3 in general don’t have removable OLPFs.

In general, a lens doesn’t show dramatically different characteristics depending on the camera where it is used with higher-end sensors in both of them. If it is soft on one camera, it is soft on other cameras, apart from some factors like differences from image circle vs. sensor size and resolution, bad back focus, in-camera processing, etc.

For Canon’s dual fisheye lenses specifically, they use only half of the sensor for each eye, so this already lowers per-eye resolution by half, compared to a stereo rig with a full sensor and lens for each eye. It is made for Canon’s sensor size, not VV, therefore the R5C catches a few more pixels horizontally at 8k DCI than the wider V-Raptor sensor at the same horizontal resolution. And even more so with the 3.9 mm dual fisheye instead of the 5.2 mm (that wouldn’t be so relevant with the R5C though). But a lower resolution image isn’t the same as a softer image unless softening is added while upscaling the image to exactly match the resolutions for both cameras.

Canon nowadays obviously also relies a lot on computational image corrections to compensate optical weaknesses for their lenses. I haven’t got my hands on such a dual fisheye for checking out what the unprocessed footage would look like, but I’d assume that it needs quite a lot of processing to make it VR-usable, just for the fact that these are fisheyes. So, if you compare the in-camera corrected footage from an R5C to RED RAW as-is out from a V-Raptor, that’s a bit unfair, a valid test will apply similar processing to the RED footage and then I doubt that there’s a notable difference in sharpness.