rerun119 you know what else strikes me super odd about all of this, is that the whole scaling issue with the canon camera is being blamed on it's IPD which is total BS to a certain point. If the IPD alone was the issue then how come footage from other consumer cameras with a much smaller IPD dont look even worse??? Think of the EVO and the VUZE cameras, those IPD distances are probably almost half of the canon system and do things looks almost twice as big? I dont think so, ive seen plenty of Vuze footage, some old scenes from other studios on this site where even shot on that, I also owned the EVO and scaling was never an issue. There is a balance between IPD, distance to camera, lens FOV and FOV rendered. I'm taking a wild guess here but I think the canon marketing is probably rounding up to 190 FOV for their lens, its probably a little lower than that, so when their software converts to 180 its actually zooming in a little more than it should, which is why the scaling is slightly larger than it should. They should be able to fix this with Mistika and not crop in as tight as the canon tool does, I just dont have the tools to precisely measure and prove this but they should. Again, it's just my guess but I'm pretty sure the canon lens is not actually a full 190 lens.

      metichemsi Hey, we actually never used rf52 projection for releases (it turned out unusable), we prepare footage for fisheye190 projection, so you have to prepare it like you would for fisheye180:

      • stereo correction
      • export fisheye circles next to each other in 2:1 aspect ratio

      I will prepare you a sample from your file you sent, fisheye190 with same scale, need to find time, probably on Monday.

        5 days later

        metichemsi Sent you the file if you care to check, let me know how is the scale, thanks.

          Hey!

          Here is an open invite for everyone to check, this samples are prepared with Canon Utility and then same thing with our workflow in Mistika.

          Download

          Personally I hardly notice any difference in scale between them, while overall scale seems big for both of them.
          So my conclusion is: there is no magic we can do here to somehow fix lower IPD, thou I am still open for debate and samples.

          I will add, that we found some deviation from our workflow in some videos and we are adding steps to avoid sub optimal quality going forward.

            Hunky-Dory I left everything without color correction so there wont be any bias based on image colors. Just increase the contrast if its so hard to watch washed out videos.

              Sandi_SLR all the fisheye footage is in a 2:1 aspect ratio. There is no sample file of the footage left in its original resolution and aspect ratio. Can you provide another fisheye sample with the original resolution and aspect ratio? That is the format which has correct scaling. its actually 256:135 aspect ratio or 8192x4320 resolution, not 8000x4000

              Thanks

                  metichemsi I provided the files as they are streamed from our platform, please stop mentioning canon original aspect ratio, there are no videos in that layout on slr.
                  Please play the files in the app and compare scale of actress between them.
                  Do you see any difference?

                      Sandi_SLR Because many VR scenes are "too big" in scaling for me I figured I'd check these out, very interested to see what comes from this...

                      With the four files I notice little (if any) difference in scale, with overall scale too big for all clips like you said.

                      But from what I understand I think metichemsi specifically meant to try exporting in the native AR because 8000/4000 is 2:1 but 8192/4320 is 1.896:1 (or 256:135).
                      Although I think the end resolution shouldn't matter as long as the original AR is used to create the actual fisheye projection, are we sure that whatever tool you use keeps that original AR for this and doesn't change/stretch it to the 2:1 end result?

                      👇️

                      metichemsi please make sure to also recreate exporting the fisheye footage from mistika with the same resolution and aspect ratio as the CRM files, its not natively a 2:1 aspect ratio. I simply drop the CRM footage on premiere and export the fisheye footage in its native resolution and aspect ratio and your r5c camera profile on DeoVR shows great scaling on that footage.

                      metichemsi I'd be very interested to compare a clip in 256:135 and 2:1 AR as well if you have those, just to see if difference in exporting like that has a noticeable effect.

                              Aerowen half of the samples are made with Canon Utility and as you noticed there is no difference in perceived scale, so why are we still talking about ratio, if you are suggesting aspect ratio matters, is the Utility working wrong too?
                              The fisheyes itself, on canon raw, are in 1:1 ratio, they are just saved on canvas of 8192x4320.

                                  Sandi_SLR why don't we post some fish eye samples si the guys can do whatever they want with these

                                    Sandi_SLR The fisheyes itself, on canon raw, are in 1:1 ratio, they are just saved on canvas of 8192x4320.

                                    Ah so I assume the actual image AR does not get changed then, but it doesn't crop either? Because I would expect that to effectively zoom in somewhat...
                                    I'm just going off what metichemsi was saying there's a difference and suspects the lens is not actually a full 190. It's possible if that's the case when converting it to 180 it zooms in more than it should. But I'll leave it to him to explain that, I have no actual experience with the cameras or the software. Just saying it sounds plausible.

                                          I'm looking back at the school class room scene with Blake and sky in pico 4, compared to the quest 2 in the pico 4 it looks fine, I put it in pt and it matched up pretty good with my background 🤔, maybe just depends what headset your using, maybe that's why @LordCrash was OK with the scale, cuz I think has bigger fov with the headset he uses...

                                          Ima look at sum more of those "big scaled scenes"

                                          But right of bat I didn't get that feeling like "whoa why it looks so big" using the pico4 🤷‍♂️

                                          rerun119 He shared a file with me that he recorded, at home, with canon, I created the same sample for him as I posted up there with fisheye190, he is free to share my sample if you guys want another comparison.

                                            We will supply more sample from other videos in coming days so stay tuned, we would really like to get to bottom of this.

                                            I will also follow up with more technical details on what is going on, to shed more light how stereo affects scale and why solving low IPD is not possible.

                                            I'm sincerely puzzled it takes so much effort for camera makers, Mistika people and all the community and SLR team to figure out some basic geometry optics.

                                              Sandi_SLR there is no difference because you are not taking into account the original resolution and aspect ratio. I am done trying to explain this to you, you are very stubborn. I provided you samples of raw footage that you can simply view in any of your fisheye profiles, original resolution and aspect ratio. Scaling is perfectly fine. If you are going to keep ignoring that then I am finished trying to explain this to you.