Probem with dissolve not black in resolve version 15 and 16

Hi,
Currently on an HDR workflow for stream, I just noticed a big problem that I have not seen for 7 years of use of Resolve.
All my AAF Avid or XML recovery dissolve are bad in Resolve (Dissolve beginning, end or inter plan)

Handing :
Take a fairly bright and contrasting shoot and place a Cross Dissolve to make a fade at the beginning of the shot.
Put the resolve in Color Management to see more defect.
Input Color Space Bypass
Timeline Color Space P3-D65 ST2084
Output Color Space P3-D65 ST2084 1000 nit

Even on an SDR screen you can see that the first image of your shoot/program is not black.

Avid and Premiere make a black image during a fade. I've never seen this before but with the HDR it amplifies the defect that is also present in SDR

I noticed this defect because the cross dissolve between the HDR Resolve and HDR Avid shoot are not at all the same in colorimetry while everything else of the program matches perfectly. Resolve seemed to me much less good in colorimetry than Avid. In the end, I go at the beginning of the program and surprise the first image of Resolve is not black.

Other tests:
Test in ACES No input and output P3-D65 ST2084 1000 nits - same problem.
Testing in Resolve version 15.3.1, 16.1.2 and 16.2 gives the same problem.
Test with a deep to color same problem.

Thank you for quickly correcting this problem on a new version.
Boissard Michael

the first image of your shoot/program is not black.

In my experience, that's normal in Resolve. The fade begins working on the first frame.

On a secondary note, it's...odd...to start a program at the very beginning of a timeline. Starting with black is more common.

Why the first or last image of a fade in Avid, Premiere and Final cut is black (value 0 in fade).
The same value on Resolve does not give a black image.
The image is not black at the end of the program, if you do not add an image without a program before the end credits.
I find it crazy.

Dissolve with cursor at the top of the shot is good for create black frame in end or start shot.
Dip to color dissolve Black is not good for the first frame because is not at 0.

No one has an idea for this problem.

I would just say that cross-dissolve is not well-defined for a single-clip operation. It is well-defined as a transition between two clips. To achieve just a fade, use the fade handles.

For an AAf conformation, I don't see myself redoing all the fades because Resolve handles them badly. Fades between shots aren't good either. On 4 software, it is the only one to give a different colorimetry at the beginning, end and inter-shots.

I agree this can be a pain when dealing with black as you always have to adjust the fade and add a 1 frame gap in the timeline if you are conforming a timeline from another software and want an exact match. Especially on the end, if you fading to black then it must end on a black frame.

In the meantime when conforming I'd suggest going with the feel of the transition - if you can't tell it's different from the original on playback, then just leave it as is.

Yes
There is a trouble with the first frame from a cross dissolve at the beginning of clip.
It seems it is the ONLY transition to have this trouble !!

Here is the Resolve project but the concern is not related to the project because I redid on several stations and with other people.
https://drive.google.com/file/d/1LIsBdrAEy9gvMtlNnzZBMzw_9VsUD5ZT/view?usp=sharing

V1 Native Confo (Import AAF mounting Avid)
V2 Render Resolve 1
V3 Render Resolve 2
V4 Render Avid (Before rountrip Avid)
V5 Render Premiere (Before rountrip Premiere)
V6 Render FCP (Before rountrip FCP)

Warning :
"Check in" About DaVinci Resolve "and make sure you have build 055 and not build 054 installed.

If you have 054 installed, re-download Resolve from the BMD web site and install that.

In this specific case, you don't need to uninstall Resolve first, Build 055 version will install over top of the 054 version without uninstalling."
Dwaine Maggart
Blackmagic Design

New tests:
1) Dip To Color Dissolve
Dip to color black dissolve between two shots.
Odd number of image dissolve (11 images) = An image of black at the end of the foreground.
Even number of image dissolve (10 images) = No black image.

2) Video Transiton vs Fusion Transition
- Video Transition Cross Dissolve generates a non-black image, problem at the start and end of the shoot. The dissolve is bad between shots because it does not give the same colorimetry as the other softwares (Avid, Premiere FCP).

Fusion Transition Cross Dissolve is perfect in black at the start and end of the shoot.
The dissolve is perfect between shots and gives the same colorimetry as the other softwares (Avid, Premiere FCP).

How to make an AAF confo and automatically resume Cross Dissolve Fusion instead of Video ?

After

If it were me, I would put a few frames of black on a video layer above, and then dissolve into program from that.

This corresponds to an alternative solution which does not correct the bug in the Resolve. I know of alternative solutions but I would like to correct the problem. I have no time to lose during the confo of 3 x 52min per day in AAF AVID rountrip.

Blackmagic does not understand the problem in the ticket I opened. If we could be several to test and exchange on the subject, it would make things happen. I just tested on Nucoda and this is identical to Avid, Premier and FCP. The other software all make an image of black at the beginning or at the end of the dissolve (gray on Resolve) and their fades between two shots are all identical (different on Resolve). This defect is negligible in SDR but impossible to ignore in HDR. I will test on Baselight shortly.

I compare with the other software, because when I grade the shoots on Resolve with the dissolve created by the conformation of the AVID AAF and I send the grading shoots (RENDER AVID AAF) in AVID to make the final render (because AVID effect) I no longer have the same colorimetry on the dissolve. I can't do rountrip with AVID in HDR. The shoots grading in P3D65 ST2084 1000nits are identical to the Resolve except in the places where there are dissolves because badly interpreted in the Resolve.

Dissolves won't affect grades when using Individual Clips renders, I don't see the problem if all transitions will be done in Avid in the end.

I am in individual clip for rendering, but when I put the grade clips in Avid the dissolve of the basic montage are reapplied on the shoots and do not correspond to those seen in Resolve. The colorimetry of the dissolve generated by Resolve during the AAF confo is not good (idem without AAF and only applied dissolve).
This causes problems between a render of Resolve and a return of the grade clips in Avid and are rendered. The Avid render dissolve to black at the start and end of shots (not the case with Resolve).
Same for the fades between shots.

Avid rendering has transitions identical to other software (FCP, Première and Nucoda). You will find my project in the discussion which demonstrates the difference in rendering between Resolve and 3 other software that uses the Resolve gradind.
You can even see the conformation on the natives which generates bad dissolves.
Thank you for your help

Blackmagic's response to my tickets

"Hi Michael,

Thanks for getting back to me.

Please be advised that after testing here further, I believe I have been able to replicate the behaviour you have seen more accurately.
However, I believe that the origin of this is due to the gamma setting that you currently have your timeline setting set to in Resolve's colour management.

As such, you may wish to test by changing this to an alternative curve such as Gamma 2.4.
Please be advised, this function will change the ways in which the tools work in Resolve, such as the cross dissolves, however should not affect the delivered image.

In regard to the additional behaviours you have reported, I have a some information for you below.

After testing here, I believe I have been able to replicate the behaviour you have reported when working with the Dip To Colour transition.
As such, I have passed information on this over to our development team for further investigation.
Going forwards in regards to this as a workaround, please be advised you should be able to change the transition curve to match what you are expecting to see.

If you have any further questions, please let me know.

Kind regards,

Xxxxxxxxxxxxx Xxxxxxx
Resolve Specialist
Blackmagic Design EMEA"

My response to the blackmagic ticket on March 23, 2020 :
Hi Xxxxxxxxx,
The problem does not come from color management, because it is present in Color Science ACES.

I am in the HDR color workflow recommended by Netflix / Amazon for Resolve. I have to do HDR in PQ and not in gamma 2.4. It was when switching to HDR with an X300 that I noticed this problem.

The problem is present in all color science. Without color management in DaVinci YRGB with timeline in Rec.709 and Gamma 2.4 the problem is present but almost imperceptible to the eye. The scope will help you display the levels and demonstrate that the first or last image dissolve is not black.

The problem does not come from color management !

How do you explain :
1. Dip To Color Dissolve
Dip to color black dissolve between two shots.
Odd number of image dissolve (11 images) = An image of black at the end of the foreground.
Even number of image dissolve (10 images) = No black image.

2. Video Transiton vs Fusion Transition
Video Transition Cross Dissolve generates a non-black image, problem at the start and end of the shoot. The dissolve is bad between shots because it does not give the same colorimetry as the other softwares (Avid, Premiere FCP).

Fusion Transition Cross Dissolve is perfect in black at the start and end of the shoot.
The dissolve is perfect between shots and gives the same colorimetry as the other softwares (Avid, Premiere FCP).

3. Dip to color Dissolve vs black frame with cursor shot
Video Transition Cross Dissolve generates a non-black image, problem at the start and end of the shoot.

Dissolve is ok with black frame with cursor shot.

4. Color Management
Test in Color science ACEScc
ACES input Device Transform : No Input Transform
ACES Output Device Transform : P3-D65 ST2084 (1000 nits)
It is the Netflix workflow

Video Transition Cross Dissolve generates a non-black image, problem at the start and end of the shoot. The dissolve is bad between shots because it does not give the same colorimetry as the other softwares (Avid, Premiere FCP).

5. Resolve version
Testing in Resolve version 15.3.1, 16.1.2 and 16.2 gives the same problem.

Thanks for your help
Kind regards,
Michael Boissard
https://www.linkedin.com/in/michael-boissard/

Responsable laboratoire, ingénieur process et workflow
Groupe Eliote
66 route de la Reine
92100 Boulogne-Billancourt

Blackmagic's response to my tickets
"Hi Michael,

Thanks for getting back to me.
Please accept my apologies for the delay in my response.

Going forwards, I will now pass information on this case over to our development team for further consideration.
As soon as I have heard back from them, I would be happy to advise you further.

If you have any further questions, please let me know.

Kind regards,

XXXXXX XXXXXX"

Great, blackmagic is investigating the problem.

I realize this was long ago, but did BM ever address this issue?

In the current version of Resolve 18 I cannot get a clean fade up from black/dissolve from black. I posted about this problem before and it was never resolved. The first frame or sometimes the first few frames have a weird bump in luminance before dropping back to nearly black. It's maddening.

That's...not normal Chris.

Can you show us? Share a small clip we can advance through frame by frame?

Or better yet, create a small Archive to share so we can examine Project Settings and such as well.

Hi Jim--How can I share a clip with you privately? I can generate an archive tomorrow; too busy today.

Thank you!
Chris

When you're ready, generate the Archive, upload to a file sharing site, post a link.

Apologies for the delays: in creating a new small project to send you, I think I figured out the issue. The weird opacity/dissolve issue only seems to occur with clips with high contrast--specifically if there are areas that are very dark and very bright adjacent to each other.

The clip fades in properly--its brightness increases evenly when viewed on the waveform frame by frame--but because the dark areas brighten apparently more slowly it seems like an odd luminance bump is occurring as the bright area brightens more quickly leaving what looks like a dark shadow in the dark areas. Not sure if this makes sense.

If I keep having the problem I'll follow up.

Thank you!