Welcome to our community

Be a part of something great, join today!

Register Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

IPP2 2021 - New Enhanced Demosaic

Awesome stuff Graeme and RED team, thanks for continuing to push further improvements!
 
Following up on my previous post, I want to confirm that I can now verify Phil's results. There were some configuration issues in my previous test, including an odd green cast to RedCine-X when GPU Processing was turned on. Not sure what's happening here with a Windows10 machine and an RTX3090 GPU with the latest nVidia drivers. At any rate, I, now, can see the magenta fringes in Davinci Resolve, but, not in RedCine-X. Yay!!
 
Graeme Nattress said:
DEB was deprecated in favour of the new CNR algorithm, which improves upon what DEB could achieve.
Click to expand...

ahh good to know.
your work is making our image acquisition easier.
You’re a rare mix of both left and right brain genius.

Phil- appreciate all of those screen grabs and analysis
of what’s going on under the hood.

Brian Timmons
BRITIM/MEDIA
 
Graeme Nattress said:
Tonal changes are not in the underlying LOG, but in the 3D LUT development. I'm just fixing up some odd precision errors in how the LUTs get interpolated, but it's not altering the underlying data or tonality, so your grade on top will work just fine and you'll be able to make it look how you want.

Graeme
Click to expand...

So, the enhanced demosaic is within the transforms? So if I use my own transform I don't get any benefits of this? Some of the things, according to Phil's images, seem to be some of the things I was aiming for when doing that transform (in detail here), but I would love to know if this works in combo with grades/LUTs done directly from the LOG image.
 
Christoffer Glans said:
So, the enhanced demosaic is within the transforms?
Click to expand...

Order of operations is CNR->Demosaic->Colorimetry calibration / colour space conversion -> log -> CDL -> 3D LUT

You can get all the demosaic benefits with any LUT on the output.

Graeme
 
Graeme Nattress said:
Order of operations is CNR->Demosaic->Colorimetry calibration / colour space conversion -> log -> CDL -> 3D LUT

You can get all the demosaic benefits with any LUT on the output.

Graeme
Click to expand...

Super awesome, now pick up that phone and call Grant Petty so we get this into Resolve asap! :encouragement:

Great work!
 
Blair S. Paulsen said:
Thanks for noting the order of operations Graeme. When some producer or post person is claiming something else I can point to this. Way more productive than a pissing match over whose assertions to follow...

Cheers - #19
Click to expand...

If you ever get questions on the image processing pipeline / order of operations, direct them to me and I'll give the authoritative answer. It's also documented in the various IPP2 papers.

Graeme
 
Yeah, simply put: This is awesome. Helium breathes new life after this update (100% crop):

4TsTkT0.jpg
 
Jacobo Martinez said:
Will this be implemented in camera or it will only be available in post? I am asking because if it isn’t available in camera, proress will definitely look different the R3D recorded in camera.
Click to expand...

In-camera demosaic is very similar to the new enhanced demosaic. However, the enhanced demosaic algorithm has some different design requirements so that it works effectively on footage that has been compressed. ProRes is always going to look a tad different (different compression scheme), but the contribution of demosaic to this is going to be negligible.

Graeme
 
Does the new Enhanced Demosaic improve CMOS smear on previous cameras/OLPFs? (I know it's completely different, but it's been an issue since Dragon in the same way that this type of fringing has been.)
 
This is awesome news and a great contribution by Greame, his team and RED. And Thanks to Phil for writing up a very nice article about how this is implemented etc. Thanks again Phil for spending some time discussing related subjects with me at this year's Cinegear Expo (Friday).
 
Mike P. said:
Does the new Enhanced Demosaic improve CMOS smear on previous cameras/OLPFs? (I know it's completely different, but it's been an issue since Dragon in the same way that this type of fringing has been.)
Click to expand...

CMOS Smear is a large area effect from the sensor, and not effected either way by demosaic.

Graeme
 
Wow, that is fantastic. How long before the new IPP2 SDK is available to other companies like Blackmagic and DaVinci Resolve? Will this new Demosaic system have a name, like IPP3?
 
Marc Wielage said:
Wow, that is fantastic. How long before the new IPP2 SDK is available to other companies like Blackmagic and DaVinci Resolve? Will this new Demosaic system have a name, like IPP3?
Click to expand...

Yes echoing this. When will this new demosaic hit Resolve? And so its easy to track, will it have a new name, as Marc suggested - like IPP3?

Congrats Graham - this looks like a pretty big deal indeed.
 
Thanks Graeme, this is the pure reason to shoot raw. Also, thanks Phil for spending all time investegating and notice us about this new feature for us. I dont use redcinex nowdays. Just Davinci. Looking forward to se this in Davinci!
 
Thanks for all the examples showing what to look for. Had a look for myself and can see the differences with the new de-mosaic. Congrats and thanks to Graeme and RED for continuing to push the image quality.
 
You must log in or register to reply here.
Back
Top