Page 2 of 2

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 1:00 pm
by rick.lang
Dmitry Shijan wrote:
rick.lang wrote:RED IPP2 version looks very dull. Are you sure you apply additional Log to rec lut node ? If you apply IPP2 n project settings it dives you the same dull result? My guess consumers and singers would like will like saturated version better if you show them doth.


Here is what I did:
1) I added your LUTs into the folder containing all LUTs used by Resolve.
2) At the project settings level, I followed your instructions.
3) I added a second and third node as instructed and added the CST effect to both nodes.

Your LUTs did not appear in the CST like all other LUTs. So to add your LUTs to the appropriate nodes:
4) I clicked on the LUT tab in the upper left bar of Resolve’s Colour Page.
5) I right-clicked on your ARRI 709 LUT and selected the menu item pop up to add it to the current node which was node 2.
6) I right-clicked on your IPP2 very soft LUT and added it to the current node which as node 3.

Did I do something wrong? Shouldn’t your LUT folder have shown as an option in the CST so that I could add the appropriate LUTs to the appropriate nodes within the CST?

I’m running Resolve 15.3.1. I know my iMac viewer quality in Resolve looked way better (better blacks and better colour) than the somewhat low key video I see on Vimeo. My iMac is set to Rec.709. The video upload to Vimeo is H.264.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 1:11 pm
by rick.lang
Wayne Steven wrote:Which one looked more like the real life one if watching in the dark for a while and similar brightness?


Hard to answer, but neither Vimeo video matched what I remember. On my computer in Resolve, the toned down version was not quite so flat as Vimeo made it.

I like the standard one better. I've seen stuff like I posted above, ts the money intended artistic choice of the concert... I range between the two depending on health but prefer the pizazz look myself.


Let’s see if Dmitry tells me I did something wrong with the toned down version. It certainly took the glare off the coloured LEDs visible in frame, but maybe I shouldn’t worry so much about that.

Remember neither version I posted has any grade applied other than the LUTs with RCM or no LUTs with ACEScct. If the no LUTs version is too strong for my taste, there’s nothing to stop me from grading the footage which I always do except for this comparison video test.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 1:22 pm
by Dmitry Shijan
Yep, two nodes... Seems you do all correct, but i still wonder why it looks so low contrast and low saturation. In my tests IPP2 is pretty well saturated, and Log3G10 to Amira709 also way more contrast looking.
Just for test can you share a single DNG frame?

On my system those LUTs are visible in Resolve LUT browser when i put them to Resolve LUTs folder

Blackmagic Pocket Camera 4K Colour Fringing and colour issue

PostPosted: Tue Apr 09, 2019 1:38 pm
by rick.lang
Can’t give you a DNG because this was recorded in ProRes 444 XQ.

I see your LUTs in Resolve’s browser on the upper left screen area but not from the drop down menu in the CST. Even after restarting Resolve. That was strange but glad my “add to current node” workaround was also correct.

Thanks for the feedback.

Aside: I may go back to CinemaDNG recordings but for this one I selected ProRes 444 XQ because it allows me to record longer. They told me two 45 minute acts so raw would have been fine but I didn’t trust their numbers so I used ProRes to extend the recording time. I was right thankfully as both acts went well over the limits for raw recording.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 2:23 pm
by Dmitry Shijan
Isn't CST menu use only Resolve native build-in presets? LUTs shouldn't be visible CST menu.

If you use ProRes check in the right click menu if input color space set to "Project"
New Resolve version have a problem - it assigns Rec709 to any ProRes source, even if it was shoot at BMDfilm Log.

Blackmagic Pocket Camera 4K Colour Fringing and colour issue

PostPosted: Tue Apr 09, 2019 3:34 pm
by rick.lang
You were right Dmitry, it was set to Rec.709 for Input Colour Space and of course that’s wrong. Set it to the project and then it used the correct Film 4.6K V3 and it has a much better look. I’ll post shortly.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 4:04 pm
by AngelDark92
Wayne Steven wrote:Which one looked more like the real life one if watching in the dark for a while and similar brightness?

I like the standard one better. I've seen stuff like I posted above, ts the money intended artistic choice of the concert. They had dual view buttons fir disks years ago, I think a dual grade button is probably needed, so they can choose a different grade. Because whatever the lighting director thinks, some people will prefers a less colorful version. This other companies issue is, that some people in the aspie range perceive colours more strongly, some dully. I range between the two depending on health but prefer the pizazz look myself.


The issue is not only related to having more or less colours unfortunately. It's just a different colour space than what we are used to. That means that some colours will even appear to be completely messed up as you can see in the video I posted above. I have also found this video showing the same issue:



Once you limit it to the correct colour space you can saturate the colours to your liking. In the previous video I posted you can also see two lights which are supposed to be different shades of red but appear to be the same colour until the correction given by the gamut limiter or the one posted by Dimitry is applied. Also the skin tone tends to go towards magenta until you use the correct colour space. I have still not had the chance to try it on BRaw though.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 4:13 pm
by Dmitry Shijan
To be honest i still can't understand what exactly gamut limiter OFX do. I check it with new gamut scope and it cuts (or compresses?) gamut in very strange and "hard clipping way". RED IPP2 gamut compression method described earlier looks way better.

Blackmagic Pocket Camera 4K Colour Fringing and colour issue

PostPosted: Tue Apr 09, 2019 4:48 pm
by rick.lang
Here’s the updated video with corrected Input Colour Space and I used h.265 10bit codec. Wayne, this seems much closer to how I remember the stage looked. This skin is so much better.



Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 4:57 pm
by deezid
Dmitry Shijan wrote:To be honest i still can't understand what exactly gamut limiter OFX do. I check it with new gamut scope and it cuts (or compresses?) gamut in very strange and "hard clipping way". RED IPP2 gamut compression method described earlier looks way better.


It does indeed. Way better handling of saturated reds and blues than anything else.
But that's why we need working alpha channels using RCM (and ACES)...

This is what happens with anything that has any kind of alpha since version 15.2.3
Screenshot from 2019-04-05 02-35-06.png
Screenshot from 2019-04-05 02-35-06.png (806.73 KiB) Viewed 8231 times

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 6:26 pm
by Dmitry Shijan
Seems on my computer alpha works without any issues. Here is imported PNG with text placed on timeline. Resolve Color management turned on. No any artifacts. Text generator looks the same. I use Resolve 16 beta.
When i switch to ACES it looks the same well. Maybe it is a GPU drivers problem or imported footage have wrong alpha settings?

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 6:30 pm
by deezid
Try using Text+
Weirdly enough enough I don't have any issue with imported 16 bit float EXR data but with transparency in fusion sequences which should have higher accuracy I guess?

Will try updating my NVIDIA driver to the latest

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 6:50 pm
by Dmitry Shijan
deezid wrote:Try using Text+
Weirdly enough enough I don't have any issue with imported 16 bit float EXR data but with transparency in fusion sequences which should have higher accuracy I guess?

Will try updating my NVIDIA driver to the latest


ok, now i see. Seems Text+ is Fusion generator with some separate internal processing color management. Probably t is just not ready or not optimized for Resolve yet.

P.S. normal text exported to EXR also have tiny black dots artifact.
ProRes444 exported from Resolve seems don't support alpha channel at all.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 7:12 pm
by deezid
Tried a few different machines now.
This problems occurse everywhere, doesn't matter if Windows, Linux, NVIDIA or AMD card...

It also applies to PNG files. They look jagged. Switching to YRGB makes everything look smooth. But yh - not an option for us ColorScience4 users. ;)

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 7:36 pm
by AngelDark92
Dmitry Shijan wrote:To be honest i still can't understand what exactly gamut limiter OFX do. I check it with new gamut scope and it cuts (or compresses?) gamut in very strange and "hard clipping way". RED IPP2 gamut compression method described earlier looks way better.


It compresses it, but all the while trying to match the colours from the bm 4 colour gamut to the rec 709 space...
Oh and by the way Davinci 16 still has the same "error" with the alpha channel on text, even when using Fusion's Text+. It's GPU accelerated so that might be the reason why.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Tue Apr 09, 2019 7:54 pm
by deezid
AngelDark92 wrote:
Dmitry Shijan wrote:To be honest i still can't understand what exactly gamut limiter OFX do. I check it with new gamut scope and it cuts (or compresses?) gamut in very strange and "hard clipping way". RED IPP2 gamut compression method described earlier looks way better.


It compresses it, but all the while trying to match the colours from the bm 4 colour gamut to the rec 709 space...
Oh and by the way Davinci 16 still has the same error with the alpha channel on text, even when using Fusion's Text+.


So how can we make developers aware of this issue introduced in version 15.2.3?
Would really love to use IPP2 gamut mapping for everything which means RCM.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 1:05 am
by Wayne Steven
AngelDark92 wrote:
Wayne Steven wrote:Which one looked more like the real life one if watching in the dark for a while and similar brightness?

I like the standard one better. I've seen stuff like I posted above, ts the money intended artistic choice of the concert. They had dual view buttons fir disks years ago, I think a dual grade button is probably needed, so they can choose a different grade. Because whatever the lighting director thinks, some people will prefers a less colorful version. This other companies issue is, that some people in the aspie range perceive colours more strongly, some dully. I range between the two depending on health but prefer the pizazz look myself.


The issue is not only related to having more or less colours unfortunately. It's just a different colour space than what we are used to. That means that some colours will even appear to be completely messed up as you can see in the video I posted above. I have also found this video showing the same issue:



Once you limit it to the correct colour space you can saturate the colours to your liking. In the previous video I posted you can also see two lights which are supposed to be different shades of red but appear to be the same colour until the correction given by the gamut limiter or the one posted by Dimitry is applied. Also the skin tone tends to go towards magenta until you use the correct colour space. I have still not had the chance to try it on BRaw though.


I was being polite AD.

Some people like the look.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 1:11 am
by Wayne Steven
rick.lang wrote:Here’s the updated video with corrected Input Colour Space and I used h.265 10bit codec. Wayne, this seems much closer to how I remember the stage looked. This skin is so much better.



Sent from my iPhone using Tapatalk


Yes. I can see more of the rosey red on the singers in the background at the end of the clip. It flashes through a bit fast, the forum still doesn't allow full screen expansion of vimeo videos on a phone.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 2:50 am
by rick.lang
Client requested two promos 30”, 2 ½’ and full video (nearly 2 hours which I completed today). Maybe I should have uploaded the longer promo. Happy to have the colour issues dealt with.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 7:30 am
by Robert Niessner
deezid wrote:
AngelDark92 wrote:
Dmitry Shijan wrote:To be honest i still can't understand what exactly gamut limiter OFX do. I check it with new gamut scope and it cuts (or compresses?) gamut in very strange and "hard clipping way". RED IPP2 gamut compression method described earlier looks way better.


It compresses it, but all the while trying to match the colours from the bm 4 colour gamut to the rec 709 space...
Oh and by the way Davinci 16 still has the same error with the alpha channel on text, even when using Fusion's Text+.


So how can we make developers aware of this issue introduced in version 15.2.3?
Would really love to use IPP2 gamut mapping for everything which means RCM.


You can submit a bug report on BMD's support website.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 4:37 pm
by Dmitry Shijan
deezid wrote:
AngelDark92 wrote:
Dmitry Shijan wrote:To be honest i still can't understand what exactly gamut limiter OFX do. I check it with new gamut scope and it cuts (or compresses?) gamut in very strange and "hard clipping way". RED IPP2 gamut compression method described earlier looks way better.


It compresses it, but all the while trying to match the colours from the bm 4 colour gamut to the rec 709 space...
Oh and by the way Davinci 16 still has the same error with the alpha channel on text, even when using Fusion's Text+.


So how can we make developers aware of this issue introduced in version 15.2.3?
Would really love to use IPP2 gamut mapping for everything which means RCM.


Not a big problem. You can use IPP2 gamut mapping in non RCM workflow as well.
1. Set timeline to YRGB (non color managed)
2. Set timeline to RedWideGamut/REDlog3g10
3. Apply CST node to every clip, set input gamma/gamut as BMDfilm (or other camera you use), set output gamma/gamut to Timeline.
(Not sure, but probably at this step you better set gamut and gamma mapping to none to keep conversion math simple)
4. Add one more node to the timeline and add RED IPP2 LUT or my custom made contrast + IPP2 color only LUTs from earlier posts.

Actually non color managed workflow may produce better colors because it use build-in input color profile and produce more vivid and calibrated colors out the box. The difference is very well visible now in new Gamut scope tool.
Also enjoy working Fusion's Text+

See the difference. BMMCC sample, non color managed at the left:

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 5:54 pm
by deezid
This workaround kinda does the job. Thanks.
Yellows turn a bit green here though and overexposed reds looks slightly jagged but with some sat mapping turned on in the first CST node it's gone. :)

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 6:46 pm
by Dmitry Shijan
This is BRAW sample with timeline set to P4K gamut and RedWideGamut. With new tools we can see that P4K gamut is just not large enough to fit all extreme saturated colors data without clipping.
Sure saturation mapping CST and Gamut limiter node tools may compress colors and fit them to smaller color gamuts but result looks less smooth.

Image
Image

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 7:01 pm
by deezid
Dmitry Shijan wrote:This is BRAW sample with timeline set to P4K gamut and RedWideGamut. With new tools we can see that P4K gamut is just not large enough to fit all extreme saturated colors data without clipping.
Sure saturation mapping CST and Gamut limiter node tools may compress colors and fit them to smaller color gamuts but result looks less smooth.


Sure. P4K gamut is kinda garbage atm. RedWideGamut fixes almost everything, but with Saturation mapping on the P4K to RedWideGamut CST node enabled, even the harshest reds look smooth after RedWideGamut conversion.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 7:08 pm
by deezid
Examples

Clip: P4K to RedWideGamut CST, Grading in RedWideGamut/Log3G10
Timeline: RedLog3G10 to Rec709 Gamma + Lum Mappaing CST, RedWideGamut to Rec709 Gamut lut
Screenshot from 2019-04-10 21-07-38.png
Screenshot from 2019-04-10 21-07-38.png (878.98 KiB) Viewed 8104 times


Clip P4K to RedWideGamut + Sat Mapping CST, Grading in RedWideGamut/Log3G10
Timeline: RedLog3G10 to Rec709 Gamma + Lum Mappaing CST, RedWideGamut to Rec709 Gamut lut
Screenshot from 2019-04-10 21-04-46.png
Screenshot from 2019-04-10 21-04-46.png (876.24 KiB) Viewed 8104 times

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 7:16 pm
by deezid
And now just for reference P4K colorspace + Extended lut in all its glory :mrgreen:

Extended Lut
Extended Lut.png
Extended Lut.png (743.21 KiB) Viewed 8101 times


RedWideGamut/Log3G10 + Grading
Screenshot from 2019-04-10 21-12-48.png
Screenshot from 2019-04-10 21-12-48.png (611.71 KiB) Viewed 8101 times

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 7:25 pm
by Dmitry Shijan
The problem is that to avoid further clipping during grading you need do a grade in RedWidegamut, and if you apply CST saturation mapping before your grades you irrevocably change colors and lowers quality of your source video. My opinion that saturation mapping is kind of destructive process and should be applied only in the end. But it need more testing. If it works for you, it may be useful as well.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 7:30 pm
by deezid
Dmitry Shijan wrote:The problem is that to avoid further clipping during grading you need do a grade in RedWidegamut, and if you apply CST saturation mapping before your grades you irrevocably change colors and lowers quality of your source video. My opinion that saturation mapping is kind of destructive process and should be applied only in the end. But it need more testing. If it works for you, it may be useful as well.


I actually see these artifacts in almost every shot with strong red lights. They're gone by either using Sat Mapping in the first conversion CST or by working inside RCM.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 7:35 pm
by Dmitry Shijan
deezid wrote:
Dmitry Shijan wrote:The problem is that to avoid further clipping during grading you need do a grade in RedWidegamut, and if you apply CST saturation mapping before your grades you irrevocably change colors and lowers quality of your source video. My opinion that saturation mapping is kind of destructive process and should be applied only in the end. But it need more testing. If it works for you, it may be useful as well.


I actually see these artifacts in almost every shot with strong red lights. They're gone by either using Sat Mapping in the first conversion CST or by working inside RCM.


Can you post dng frame?

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 7:44 pm
by deezid
Dmitry Shijan wrote:
deezid wrote:
Dmitry Shijan wrote:The problem is that to avoid further clipping during grading you need do a grade in RedWidegamut, and if you apply CST saturation mapping before your grades you irrevocably change colors and lowers quality of your source video. My opinion that saturation mapping is kind of destructive process and should be applied only in the end. But it need more testing. If it works for you, it may be useful as well.


I actually see these artifacts in almost every shot with strong red lights. They're gone by either using Sat Mapping in the first conversion CST or by working inside RCM.


Can you post dng frame?


That was from a Braw test.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 8:05 pm
by Dmitry Shijan
Also make sure your RAW and CST settings match each to other.

Here is example how we can switch between ColorScience v1 and v4 with older BMMCC camera (i guess):

Image

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 8:18 pm
by deezid
Dmitry Shijan wrote:Also make sure your RAW and CST settings match each to other.


Already did. When using Braw only Gen4 is possible anyway.

But inside an YRGB timeline it needs Sat mapping on the first CST conversion node to be 100% clean in highlights.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 8:28 pm
by deezid
Another observation. When recording Braw you can use Rec2020Gamut/Rec2100ST2084PQ as your colorspace. After converting to RedWideGamut/RedLog3G10 it seems to be even cleaner in red highlights :)

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 8:40 pm
by Dmitry Shijan
Interesting. You can also try to play with gamma settings in BRAW and CST. Seems BRAW allows more useful options to stars than DNG.
Current workflow looks like :
Sensor Linear -> P4kBMDflm -> CST P4kBMDflm to RedLog3g10 -> IPP2 LUT

I guess if you change P4kBMDflm to Linear in BRAW tab you simplify workflow and bypass 4kBMDflm gamma which may be also not too wide enough and produce clipping.
So new workflow will look like :
Sensor Linear -> CST Linear to RedLog3g10 -> IPP2 LUT

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 8:48 pm
by deezid
Dmitry Shijan wrote:Interesting. You can also try to play with gamma settings in BRAW and CST. Seems BRAW allows more useful options to stars than DNG.
Current workflow looks like :
Sensor Linear -> P4kBMDflm -> CST P4kBMDflm to RedLog3g10 -> IPP2 LUT

I guess if you change P4kBMDflm to Linear in BRAW tab you simplify workflow and bypass 4kBMDflm gamma which may be also not too wide enough and produce clipping.
So new workflow will look like :
Sensor Linear -> CST Linear to RedLog3g10 -> IPP2 LUT


Tried linear gamma but results in more artifacts again.
Rec2020 Gamut + Rec2100 PQ Secene gamma seems to be the cleanest

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Wed Apr 10, 2019 9:41 pm
by deezid
Nice addition in 16 is, you can make use of adjustment clips to entirely bypass the gamma/gamut conversion for any rec709, text, alpha, fusion etc... layers

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Thu Apr 11, 2019 5:38 pm
by rick.lang
Heads-up. I have noticed in the course of working on that recent concert, that I’m getting ugly artifact stair-casing on the arm of one of the backup singers whenever a strong red light hits her arm against a very dark background. It only has a tiny hint when using ACES in h.264. But bad using Resolve and RCM in h.265. When I changed to RCM in h.264, it’s good again. I’m quite surprised that HEVC has that severe a problem and I’m going to stay with h.264.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Apr 12, 2019 7:18 am
by Wayne Steven
Could it be magnifying something left over in the debayering process? If so, the left over should turn up a little in other rendered codecs, but h265's aggressive compression has magnified it. It would probably be very minute under magnification.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Apr 12, 2019 1:46 pm
by rick.lang
It might be as if I looked hard in h.264 there is a little flaw visible, but h.265 has gone ugly on it. Only on that bright curved edge against the very dark background. I did also like the increase in pleasing colour that h.264 8bit gave me over the slightly anemic colour from h.265 10bit colour.

In theory the 10bit colour should have been better than 8bit, but it’s not! I’ve disliked HEVC before when I found it exaggerating other defects so I’m just going to leave it alone until I hear it’s been improved. I was really looking forward to it a year before it was available. But the only good thing I have to say about it is it does live up to the expectations in savings but at the cost of some lost colour and exaggerated imperfections.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Apr 12, 2019 2:14 pm
by John Paines
rick.lang wrote:but h.265 has gone ugly on it.


Have you tried it outside Resolve? Handbrake, for example, via Prores?

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Apr 12, 2019 2:46 pm
by Wayne Steven
That's a good point. The render might be playing a part. I have a favourite pocket camera and you look at the screen and bird footage, I see off colour and flashing blocks, and I think this camera's had it. But when I go and view the footage outside the camera, the birds are well shaped, and the colour is good.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Apr 12, 2019 2:50 pm
by Wayne Steven
Rick, the h265 representation is probably misaligned or something. Those 8 bits, if they where incorrectly spread across the ten bit, gamma curve, colour space, or simply the HDR curves are wrong, it will affect something. On the other end is the h264 8 bit exaggerated?

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Apr 12, 2019 3:45 pm
by rick.lang
John Paines wrote:
rick.lang wrote:but h.265 has gone ugly on it.


Have you tried it outside Resolve? Handbrake, for example, via Prores?


John, I have not tried that. Good idea to try; if HEVC works well externally, then there’s an issue within the Resolve implementation of HEVC. I shot the concert in ProRes 444 XQ, and could Deliver the graded footage as ProRes 444 XQ and see if externally generated HEVC 10bit is best.

There are some controls on the Resolve’s Deliver page for the h.xxx codecs but the manual doesn’t explain what they do so I leave them on Auto. Someday I’ll try to get to the bottom of this but not at the moment unfortunately since h.264 is a reasonable workaround for this client.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Apr 12, 2019 3:51 pm
by rick.lang
Wayne, it feels like a render issue dealing with extreme colour intensity differences, but I’ve seen it before when it did a shot outdoors of fascia on a building where it didn’t handle the roughly horizontal edge lines that contrasted with the light coloured fascia. With my concert video, it’s making a similar mess of essentially vertical edge lines.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Sat Apr 13, 2019 12:15 am
by Wayne Steven
Its all strange. You presume people put in the effort to get it right first time, and undoubtedly Resolve is magnitudes better than android, but who knows where they got the h265 code from.

Now, I have noticed really weird artifacts (lots and lots) in demosiacing maybe, years back, it could be a flaw missed by the debayering.

Test it next time. Have you tried an olpf substitute? If you notice it is in relation to a particular pixel pattern position, it's likely related to that. As the camera moves out or into alignment with it, it will change. One of the problems with no olpf on Bayer.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Sat Apr 13, 2019 4:21 am
by rick.lang
Does nothing to feel like that’s a plausible cause as it appears several times in the video and I’m always panning and tilting during a performance so I don’t think I’d be able to repeat a position with such precision as I would include that singer. It has to be bright light along the boundary of her arm. I watched the h.264 video playback this morning and all
Is well now.


Sent from my iPhone using Tapatalk

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Sat Apr 13, 2019 2:31 pm
by Wayne Steven
It's Bayer, that's which they use soft olp filters. A solution has to be defined with it, or if its small enough, just live with it. Bayer is a bit toxic that way, why I objected to it many years back. Unfortunately x3 was not developed to video as an alternative, and the new video version is I think a compromise.

Relevant to the repeating Bayer pattern, 2300 times across. You are right, it would be the high contrast difference between two or more Bayer pixels at whatever angle the contrast/colour differences cause it to be non neatly defined. You can define a solution for the future.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Sep 06, 2019 2:31 pm
by grex84
Dmitry Shijan,

I know this thread hasn't been replied to in a while. Does your workflow with BRAW remain the same or have there been any new developments?

Sensor Linear -> CST Linear to RedLog3g10 -> IPP2 LUT

I've tried this and for some reason i get very desaturated colors.

Can we make sure i'm doing this right?

- Project settings is Davini YRGB, not color managed
- on the BRAW clip settings i set the gamma to Linear. What about the color space?
- Add a CST node with a Linear to RedLog3g10 on the gamma. What about Color space? Tone Mapping and Sat Mapping off.
- Add another node with IPP2 or Amira LUT your provided in earlier posts.

Is that it?
Grading happens between the CST and LUT correct?

Thanks for your help on this

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Sep 06, 2019 3:28 pm
by Giles Sherwood
I honestly didn't feel that great about arbitrarily converting to/working in RED colorspace, so I tried some other wide gamut working spaces and they seemed to handle the BMP4K's super-saturated colors pretty well.

I use a CST at the beginning of my node tree to go from BMDP4KFilm Gen 4 > P3-D65 (or Rec 2020 if I'm feeling saucy), then use a second CST at the end of the node tree to go from the wide gamut working space to Rec. 709. Tweak saturation mapping on either end to taste.

I felt this gave me much better overall color and roll-off than using RED method, but IIRC in some instances of super-red LED/neon lighting you did still have a bend towards yellow--which I ultimately preferred to the somewhat washed-out way the RED colorspace rendered them, but YMMV.

If anyone with better under-the-hood knowledge than me wants to blow some holes in this workflow, I wouldn't take it personally.

Re: Blackmagic Pocket Camera 4K Colour Fringing and colour i

PostPosted: Fri Sep 06, 2019 4:42 pm
by Dmitry Shijan
grex84 wrote:Dmitry Shijan,

I know this thread hasn't been replied to in a while. Does your workflow with BRAW remain the same or have there been any new developments?

Sensor Linear -> CST Linear to RedLog3g10 -> IPP2 LUT

I've tried this and for some reason i get very desaturated colors.

Can we make sure i'm doing this right?

- Project settings is Davini YRGB, not color managed
- on the BRAW clip settings i set the gamma to Linear. What about the color space?
- Add a CST node with a Linear to RedLog3g10 on the gamma. What about Color space? Tone Mapping and Sat Mapping off.
- Add another node with IPP2 or Amira LUT your provided in earlier posts.

Is that it?
Grading happens between the CST and LUT correct?

Thanks for your help on this


See info starting from this post for workflow details viewtopic.php?f=21&t=65149&p=537852#p537852