Blackmagic Forum

When I enable "Show picker RGB value" in the Colour page viewer, it is guaranteed to crash Resolve Studio v14.3 under Windows 10.

I have had some instances in previous versions but not as constantly as this.

When I try too attach the diagnostic file (23,937KB) I get this response "413 Request Entity Too Large".

You need to put the file on a file sharing service and provide a link to it here.

Thank you for your speedy reply. Here is a link to the log zip.

https://sabercathost.com/lVoY/DaVinci-R ... 180114.zip

Tony, I just want to mention that I'm not seeing that behavior on my machine. I can show the picker values w/o causing a crash.

PeterMoretti wrote:Tony, I just want to mention that I'm not seeing that behavior on my machine. I can show the picker values w/o causing a crash.

Peter - Thank you for that. I only ever saw it very occasionally until recently and so I am wondering if it is a conflict with something on my machine - but it has definitely got worse with the latest Resolve version. However, I would not know where to begin to isolate the problem.

I am hoping that Blackmagic will work their magic on the log files and I hope that they may be able to point me in the right direction.

I don't see anything obvious in the logs. There is no hard crashing happening, as best I can tell.

In Resolve Preferences - System - Video I/O, set For Capture and Playback Use and For Resolve Live Use, both to None. Save the preferences. Restart Resolve.

Now does the problem happen? If not, try installing the current Desktop Video 10.9.10 driver and re selecting the Intensity Pro card and see if its any better or not.

Dwaine, Thank you for your guidance. I followed it and updated Desktop Video and also uninstalled and updated to the latest NVidia graphics driver.

It made no difference and the crash still happens whether or not I have the Intensity Pro selected in System Preferences.

The only thing I can add - as you will be able to see from the short attached iPad video - is that the crash happens when the cursor reached the bottom of the frame. I can replicate that consistently now.

https://sabercathost.com/9kYb/RGBPICKER_CRASH.MOV

An internet search turned up this posted on 18 Dec 2017 at:

https://forums.creativecow.net/thread/277/41013

So it is not just me, and I quote:

"LAB color nodes offer advantages and looks that are useful and appealing. Usually setting up nodes and adjusting the LAB A & B curves works well. And LAB A & B can also useful in splitting chroma (A & B) and luma (for sharpening, for example). Most of the node tree is in RGB.

And of course one common and constantly-used tool is the eye-dropper with the "Show picker RGB value" selected. But lately this -- within seconds -- reliably crashes the program. First the cursor does not change to an arrow when moved out of the viewer pane but remains as an eye-dropper. This situation means it will crash within a second or two."

To test further, I started a new project using the downloadable tutorial material for the Paul Saccone book.

The RGB picker caused the same crashes as it did with my own sources.

I can not reproduce. Do you use OFX? other settings? Project settings?
It's crazy..

Jean Claude wrote:I can not reproduce. Do you use OFX? other settings? Project settings?
It's crazy..

I am not using any OFX or even grading anything. I am currently just looking at some clips on the timeline.

I am not sure how I can comment on my settings as there are so many of them.

Maybe export your project (.drp)? hoping to reproduce other than you?
(post the .DRP on wetransfert or dropbox or ... whatever you want)

I have been experimenting with settings and seem to have found a solution in the hardware configuration.

The default is:

GPU processing mode [Auto]
(Use display GPU for compute is OFF)
Video card 1: 980Ti
Video card 2: 1080Ti

If I set it manually to:

GPU processing mode [CUDA]
GPU selection mode [Manual]
and tick on the 1080Ti
and tick off the 980Ti

then the problem with the picker disappears and no amount of wiggling of the mouse will make it crash.

That's interesting, because your logs showed the 1080Ti being used as the only GPU and the 980 being a GUI only card.

So I'm not sure why you didn't see in Preferences - Hardware Config - GPU's, the 980 being listed as a GUI only card. You shouldn't have seen the 980 in the GPU list as Video Card 1: 980Ti unless the "Use Display GPU for Compute" selection was selected.

What you've manually set seems to be what the log shows was happening anyway.

Do you have displays connected to both cards, or just the 980 card?

Dwaine Maggart wrote:That's interesting, because your logs showed the 1080Ti being used as the only GPU and the 980 being a GUI only card...
Do you have displays connected to both cards, or just the 980 card?

I can't remember and am not near the machine at present. Will check later.

I am seeing this with my iMac Pro (see signature).

Here's dropbox link to a crash log and the project: https://www.dropbox.com/sh/1jdl29htxe71 ... Bte1a?dl=0

M

The 2 crashes at 19:32 and 21:00 were caused by the color picker?

Does the crash happen on any of the 6 clips?

Dwaine Maggart wrote:That's interesting, because your logs showed the 1080Ti being used as the only GPU and the 980 being a GUI only card... Do you have displays connected to both cards, or just the 980 card?

I had two displays connected to the 980 only. Under v12 the 980 was GUI and 1080ti was compute.

So I have removed the 980 and restarted Resolve with the 1080ti only.

Use display gui for compute was unticked which seems odd for a one card machine, so I ticked it on. The rgb picker continued to crash.

I changed the setting to CUDA mode, manual and it does not crash now, I hope.

I wonder is the AUTO thing not doing its work or maybe something got corrupted. But the issue does not seem to be project specific. I have used three different projects for my tests and none have any grading, just viewing in the colour page

Dwaine Maggart wrote:The 2 crashes at 19:32 and 21:00 were caused by the color picker?

Does the crash happen on any of the 6 clips?

Was this addressed to me or other reporters? I definitely experienced crashes while attempting to use the color picker. I should add that I've started using a Decklink Duo (not 2, original) with Desktop Video 10.9.10. Not sure if my output path was one of the SDI ports or "None" but definitely trying to use color picker in GUI when it crashed.

I spoke too soon!

I returned to grading yesterday and the instant RGB picker crash was still happening.

After one instance, I decided to restart the desktop and Windows 10 brought up a screen saying that it could not restart as a program was still active. It stated that it was a:

"NvFBCPlugin Window".

The program properties of a DLL with a similar name indicate that it is a frame buffer.

Maybe there is a clue in that as to what is going on?

I have uninstalled this version with my current project deleted and started it again as I was only just laying it out for some simple grading. I am keeping my fingers crossed for today's work.

I have attached a link to the latest diagnostic file.

https://sabercathost.com/599a/DaVinci-R ... 193422.zip

If the crashes continues I shall go back to the previous version.

Thank you for your help.

This is hopeless! I can't get any work done!

I have reverted to version 14.2.1.007 but the problem persists.

I created a diagnostic log but it seems to be a day behind the current situation and looks the same as the one I made from v14.3 and uploaded earlier.

I have disabled everything in my startups except for BM CheckVersion PCI and BM Streaming Server.

I have uninstalled MalwareBytes.

The problem persists. Repeated checking shows that it crashes when the tip of the cursor reaches the very bottom line of the picture area in the Colour page viewer. I have done that very slowly and carefully quite a few times.

The Windows 10 Event log shows different modules crashing at different times even though the same fault is occurring but I wouldn't know if this is significant.

Examples are:

Faulting application name: Resolve.exe, version: 14.2.1.7, time stamp: 0x5a65bf61
Faulting module name: Resolve.exe, version: 14.2.1.7, time stamp: 0x5a65bf61
Exception code: 0xc0000005
Fault offset: 0x0000000001f2710e
Faulting process ID: 0x1fa8
Faulting application start time: 0x01d3a41e4d0d7ede
Faulting application path: C:\Program Files\Blackmagic Design\DaVinci Resolve\Resolve.exe
Faulting module path: C:\Program Files\Blackmagic Design\DaVinci Resolve\Resolve.exe
Report ID: 145ee83f-7e92-44cd-97c3-f35f4b7f998f
Faulting package full name:
Faulting package-relative application ID:

Faulting application name: Resolve.exe, version: 14.2.1.7, time stamp: 0x5a65bf61
Faulting module name: pthreadVC2.dll, version: 2.9.1.0, time stamp: 0x58255014
Exception code: 0xc0000005
Fault offset: 0x00000000000045df
Faulting process ID: 0x1e10
Faulting application start time: 0x01d3a41e16b6befc
Faulting application path: C:\Program Files\Blackmagic Design\DaVinci Resolve\Resolve.exe
Faulting module path: C:\Program Files\Blackmagic Design\DaVinci Resolve\pthreadVC2.dll
Report ID: 47b67235-91a6-44b9-be68-f1f4bf5e7eb1
Faulting package full name:
Faulting package-relative application ID:

Faulting application name: Resolve.exe, version: 14.2.1.7, time stamp: 0x5a65bf61
Faulting module name: FairlightPage.dll_unloaded, version: 1.0.0.0, time stamp: 0x5a602a14
Exception code: 0xc0000005
Fault offset: 0x00000000008ccad9
Faulting process ID: 0x1998
Faulting application start time: 0x01d3a41d6b586b7f
Faulting application path: C:\Program Files\Blackmagic Design\DaVinci Resolve\Resolve.exe
Faulting module path: FairlightPage.dll
Report ID: e2642487-51eb-48b7-aca7-7b593245db06
Faulting package full name:
Faulting package-relative application ID:

Tony, does this happen with all projects/types of source footage?

PeterMoretti wrote:Tony, does this happen with all projects/types of source footage?

Yes, as far as I can tell.

My stuff (grass valley hqx) and the sample files from the Saccone Resolve book have been tested.

Even created a new database on a different drive and created new projects.

I tried a new thing today.

I used Menu\Reset UI Layout as I had been working in Full Screen.

And there have been no crashes so far in that mode.

Changing back to full screen or dual screen brings the crashing picker issue back.

Still, I am wondering if Blackmagic have abandoned me as there has been no feedback in a while from them.

Should I move to the support email system instead?

To continue (talking to myself it would seem)!

Today, I started a new project to work on some of the material in the Saccone book.

I choose to turn on the RGB picker but nothing showed in the viewer except the mouse cursor.

Then, when I chose Menu\Workspace\Reset UI Layout, the picker appeared again and did not crash.

When I switched to Full Screen mode, the picker crashed the application.

Has BM any view on this or will they work on a fix?

Here is today's diagnostic log in case it is more informative than the previous one.

https://sabercathost.com/59cm/DaVinci-R ... 150526.zip

Tony, I just want to mention that I tried using the picker with the full screen viewer and could not replicate what's obviously happening on your system.

P.S. May I ask why you use Sabercat for file sharing? It feels much more clunky and slower than services like Dropbox, IMHO. But competition is good, so I'm glad that people are using other services.

P.P.S. Have you tried using just the 980 instead of just the 1080?

Tony Martin wrote:This is hopeless! I can't get any work done! I have reverted to version 14.2.1.007 but the problem persists.

Just as a workaround: is it possible to get the job done a different way and not use the color picker? I honestly almost never use it, and instead rely on the scopes to tell me when two colors don't match (as with a commercial product shot or something like that).

It sound like a GPU driver issue. If it was a fundamental problem we would have 1000s seeing this.
Not sure of your card use status but I would just use the 1080 and a single monitor.

Then please list the CUDA driver detail.

Peter Chamberlain wrote: If it was a fundamental problem we would have 1000s seeing this.

I don't doubt that, but at least one other user posted in this thread:

Michael Tiemann wrote on Wed Feb 07, 2018 11:49 am:
"I definitely experienced crashes while attempting to use the color picker."

Could it be related to my using the original Intensity Pro rather than the newer 4K version?

Do you still offer compatibility with the Intensity Pro (non 4K)?

My CUDA details are:

NVIDIA System Information report created on: 02/20/2018 13:01:41
System name: EDIT

[Display]
Operating System: Windows 10 Pro, 64-bit
DirectX version: 12.0
GPU processor: GeForce GTX 1080 Ti
Driver version: 390.77
Direct3D API version: 12
Direct3D feature level: 12_1
CUDA Cores: 3584
Core clock: 1480 MHz
Memory data rate: 11010 MHz
Memory interface: 352-bit
Memory bandwidth: 484.44 GB/s
Total available graphics memory: 27566 MB
Dedicated video memory: 11264 MB GDDR5X
System video memory: 0 MB
Shared system memory: 16302 MB
Video BIOS version: 86.02.39.00.90
IRQ: Not used
Bus: PCI Express x8 Gen3
Device ID: 10DE 1B06 63903842
Part Number: G611 0050

[Components]
nvui.dll 8.17.13.9077 NVIDIA User Experience Driver Component
nvxdplcy.dll 8.17.13.9077 NVIDIA User Experience Driver Component
nvxdbat.dll 8.17.13.9077 NVIDIA User Experience Driver Component
nvxdapix.dll 8.17.13.9077 NVIDIA User Experience Driver Component
NVCPL.DLL 8.17.13.9077 NVIDIA User Experience Driver Component
nvCplUIR.dll 8.1.940.0 NVIDIA Control Panel
nvCplUI.exe 8.1.940.0 NVIDIA Control Panel
nvWSSR.dll 23.21.13.9077 NVIDIA Workstation Server
nvWSS.dll 23.21.13.9077 NVIDIA Workstation Server
nvViTvSR.dll 23.21.13.9077 NVIDIA Video Server
nvViTvS.dll 23.21.13.9077 NVIDIA Video Server
NVSTVIEW.EXE 7.17.13.9077 NVIDIA 3D Vision Photo Viewer
NVSTTEST.EXE 7.17.13.9077 NVIDIA 3D Vision Test Application
NVSTRES.DLL 7.17.13.9077 NVIDIA 3D Vision Module
nvDispSR.dll 23.21.13.9077 NVIDIA Display Server
NVMCTRAY.DLL 23.21.13.9077 NVIDIA Media Center Library
nvDispS.dll 23.21.13.9077 NVIDIA Display Server
PhysX 09.17.0524 NVIDIA PhysX
NVCUDA.DLL 23.21.13.9077 NVIDIA CUDA 9.1.84 driver
nvGameSR.dll 23.21.13.9077 NVIDIA 3D Settings Server
nvGameS.dll 23.21.13.9077 NVIDIA 3D Settings Server

Tony, have you tried disabling the Intensity Pro?

Tony Martin wrote:

Peter Chamberlain wrote: If it was a fundamental problem we would have 1000s seeing this.

I don't doubt that, but at least one other user posted in this thread:

Michael Tiemann wrote on Wed Feb 07, 2018 11:49 am:
"I definitely experienced crashes while attempting to use the color picker."

Note that I don't have an Intensity Pro, but rather a DeskLink Duo. And I don't have (and never have had) CUDA drivers on my Radeon-based iMac Pro.

PeterMoretti wrote:Tony, have you tried disabling the Intensity Pro?

Not near machine now but AFAIR I did try that.

But, on reflection, how could it be relevant? The output goes to an external monitor and the colour picking is done on the GUI monitor.

Tony,
(perhaps)
Can you try reinstalling NVIDIA 390.77 drivers but requesting a CUSTOM installation => check ON: Perform a new installation?
(Warning: it completely resets NVIDIA settings by default)

and Reboot PC (default)

After: uninstall BMD Desktop Video => after Reboot
After: reinstall BMD Desktop Video => after Reboot
and test?

If it does not work :
download DDU and launch it (DDU is an NVIDIA or AMD / ATI cleaner)
http://www.guru3d.com/files-get/display ... oad,1.html
After => reinstall NVIDIA 390.77 drivers => reboot

and test?

The RGB picker from Davinci Resolve is very accurate.

Jean Claude wrote:Tony, (perhaps) Can you try reinstalling NVIDIA 390.77 drivers but requesting a CUSTOM installation ... [and use a cleaner] ... and test?

I have already done more or less that (with reboots at every stage) including using an nVidia cleaner in Windows Safe Mode. I am reluctant to do anything at the moment as I am working on some old 8mm film transfers. You can't really call it grading as it is so awful but you can make most of it look a bit less bad.

I use the RGB picker as an analysis tool because of the unevenness of colour and colour density across the frame and, despite always using my eyes and the vectorscope, there are still times when I want to see what exactly the problem is using the picker tool - mostly as a double-checking device.

At the moment, I am only using it in extremis and, since updating to the latest BM desktop video software, I have avoided crashes. But whether that is a real solution or whether it is because I don't sweep the picker any more but only read an exact spot and switch it off again. It's tedious but safe.

So I don't really know where we are at with this. I have mailed BM Support and I await their findings.

Thank you all for your help and encouragement.

Jean Claude wrote:The RGB picker from Davinci Resolve is very accurate.

RGB_PICKER_LEVELS.jpg

Jean Claude, can you post a link to those color bars?

Also are you sure the file isn't being interpreted as video levels instead of full levels (which it should be)?

Tony Martin wrote:I use the RGB picker as an analysis tool because of the unevenness of colour and colour density across the frame and, despite always using my eyes and the vectorscope, there are still times when I want to see what exactly the problem is using the picker tool - mostly as a double-checking device.

I'm not trying to be combative, but I'm telling you that many, many experienced colorists get along just fine with just the scopes and accurate, calibrated monitors. It's extremely rare that I ever use the color picker with the on-screen readouts because there are so many perceptual factors involved with matching. To me, it's got to match in motion, with the edit, and that's actually more important than stills per se.

I'm a big fan of matching with Gallery stills and just flipping back and forth (rather than, say, a side-by-side comparison), and have been doing that kind of thing for many decades. I don't dispute there are unusual situations where you might need an absolute pixel match, and I also agree that this should not crash the program.

PeterMoretti wrote:

Jean Claude wrote:The RGB picker from Davinci Resolve is very accurate.

RGB_PICKER_LEVELS.jpg

Jean Claude, can you post a link to those color bars?

Also are you sure the file isn't being interpreted as video levels instead of full levels (which it should be)?

Peter,

The link to the chart is:
https://www.belle-nuit.com/test-chart

My test (above) is in Full (vs video level)
Here the same chart in full vs video level with the DCTL of Paul Dore which turns the pixels = 0 in red
(no Lut for the test)



Edit : and image FULL after export TIFF converted to jpg

Jean Claude, I think maybe I'm missing something or misunderstanding something.

The color picker seems perfectly accurate. I thought in the chart the 255, 255, 255 swatch that the picker was parked on was supposed to be 255, 251, 239 with alpha opacity of 231. But that's actually not the case, it's 255, 255, 255.

I double checked with Photoshop. Each of those numbers printed on the chart each describes a different chip of diminishing brightness. so 251 is 251, 251, 251.

So I think, like I said, I'm misunderstanding your post.

I'm also seeing crashes when using the RGB picker. Resolve Studio v14.3.0.014, Windows 10 Pro version 1709, dual Nvidia GTX TITAN X with driver v390.77.

Like Tony, I can reproduce it by moving the picker outside the video (but still within the viewer, meaning the video is zoomed out so it's smaller than the viewer).

roger.magnusson wrote:I'm also seeing crashes when using the RGB picker. Resolve Studio v14.3.0.014, Windows 10 Pro version 1709, dual Nvidia GTX TITAN X with driver v390.77.

Like Tony, I can reproduce it by moving the picker outside the video (but still within the viewer, meaning the video is zoomed out so it's smaller than the viewer).

I too was running nVidia v 390.77.

However, as I noted above, the issue seems to have improved (disappeared? - only time will tell) when I installed Backmagic Desktop Video v10.9.11 when I had previously been on v10.9.10.

Whether that would work for you or not, I don't know.

It's really is all black magic!

I'm already on BMD Desktop Video v10.9.11, using a DeckLink 4K Pro.

Using Resolve studio full 14.3.0.014, Win 64 Pro, 64 GB Ram, Dual GHT1080Ti cards, 6 core I7 machine.

I have to concur, there is indeed a severe "problem" with the the Set White Balance picker, and the Set White Point and Set Black Point pickers on the Colors Page, near the colors wheels.

I've tried on several different occurances, several different clips, all H.264, 8 bit, or 10 bit, and I have fatal crashes more often than not if I try to use any of these three tools.

I thought I'd give em a try, but they are for now a "do not touch" tool for me in the DR app.

There IS a problem BMD.

I'm experiencing the same problem in both 14.3 and 15 beta (Studio) and I don't have any BMD hardware - just an Intel system with an NVIDIA 1070 (by NVIDIA) card. Every time I see that color picker on-screen it makes me cringe. Resolve crashes and sometimes the color picker doesn't place a point/points(s) on the curve when I click in various places on the image and sometimes the values it returns are garbage or it just doesn't read them out at all (this happened I think when I had a LOG lut applied to the clip).

MikeMeagher wrote:I've tried on several different occurances, several different clips, all H.264, 8 bit, or 10 bit, and I have fatal crashes more often than not if I try to use any of these three tools.

What happens if you convert the H.264 files to something like DNxHD or DNxHR and then use those transcoded files instead? I'm not seeing the problem on the Mac at all, but I almost never use these features. To me, there are too many advantages doing it all manually. Scopes will give you all the information you need about color and levels... unless you have an unusual situation like having to specifically match a CMYK print value or something. Even then, I would just import the image, do a split screen, and match it by eye.

I concur that there are other more precise ways to adjust...but my reason in reporting was to point out that these features appear to have a bug at times so BMD can fix em for those who choose to use them. I sometimes in some cases want to adjust white and black point to a slate image...for a quick n dirty adjustment.

MikeMeagher wrote:I concur that there are other more precise ways to adjust...but my reason in reporting was to point out that these features appear to have a bug at times so BMD can fix em for those who choose to use them. I sometimes in some cases want to adjust white and black point to a slate image...for a quick n dirty adjustment.

Your crash logs might help Mike.