Creative Communities of the World Forums

The peer to peer support community for media production professionals.

Activity Forums DaVinci Resolve A very interesting R3D test – good results

  • A very interesting R3D test – good results

    Posted by Blase Theodore on July 20, 2011 at 3:17 am

    I’ve always been a big fan of going back to the debayer to get the best possible exposure out of an R3d before I start tweaking with primary nodes. Aside from getting my temp and tint right, I’d tweak both the shadow and the ISO to expand the image to a full curve.

    I did this because I assumed being closest to the raw data would get me the best image, and I’d have to stretch it less post-debayer.

    Turns out this isn’t the best approach.

    I took a bunch of shots that came out of the camera pretty flat. (Pre-MX chip, lifted shadows, low highlights.)

    I did 2 tests:
    In test 1, I left them at very low ISO’s (like 200), and didn’t touch shadows in they debayer. And I stretched them waaay out with a primary node.

    In test 2, I pre-tweaked the debayer to get to the same place I had gotten with my primary correction.
    I assumed test 2 would prove better results.

    I set render to full quality premium, 16 bit, and rendered out several shots. Turns out they were basically indentical. In fact on a really dark shot, Test 1 actually turned out much nicer than test 2.

    From this I think the following is possibly true:
    – that 16 bit debayer access means that the debayer probably isn’t a bottleneck to the raw data.
    – setting the debayer to an under-exposed ISO, and then stretching the image, gets a cleaner image.

    Cheers,
    Blase

    Blase Theodore replied 14 years, 9 months ago 4 Members · 5 Replies
  • 5 Replies
  • Vladimir Kucherov

    July 20, 2011 at 3:21 am

    I tried a similar test with near-clipped highlights and results coincide with yours – with 16 bit debayer, fussing with ISO doesn’t seem to matter all that much.

  • Joseph Mastantuono

    July 20, 2011 at 12:09 pm

    Thanks for posting this. I wouldn’t have assumed this was the case. It will make things a bit faster as messing with the red stuff on a per shot basis isn’t fun.

    Question though, Can you describe how it looks better? Less noise? Smoother curve?

    Joseph Mastantuono
    Online Editor – Colorist – Post Consultant
    Brooklyn based finishing at reasonable prices
    917.969.1583

  • Blase Theodore

    July 20, 2011 at 3:33 pm

    Interestingly the non-tweaked shot had a much lower noise profile, and a smoother curve.
    I think the raw sensor on the old RED’s was analogous to a 250 or 320 ISO, and that’s where the curve naturally wants to be.

  • Margus Voll

    July 20, 2011 at 9:25 pm

    So the MX could be left to its 800 something ?

    Margus

    https://iconstudios.eu

  • Blase Theodore

    July 20, 2011 at 10:18 pm

    I think the MX is probably closer to 400 ISO. I’d bump it down to 400 or 500 and stretch it up rather than going to 800.

We use anonymous cookies to give you the best experience we can.
Our Privacy policy | GDPR Policy