-
Difficult to fine tune colors in Resolve
-
jPo, you’re right as usually. Technically the marker which would support the development of a third type of color receptor, and tetrachromatisis, would only be found in a second X chromosome.
https://JuanSalvo.com
https://theColourSpace.com -
Itai, can you post a drp of your project. I’m curious to solve this one. 🙂
Also a screen grab of your system preferences>display>color page. Have you tried to “calibrate” the apple Cinema Display… Maybe with something like a spyder colorimeter?
https://JuanSalvo.com
https://theColourSpace.com -
Are you using a panel?
Tweaking values with numeric pad or mouse gives you this “odd jumps”.
When working with a panel, you get much more range than whats showing in the gui.Resolve Beta 10 OSX 10.8.2
Macpro 5.1 12×2,66
GTX 570 MacVidcards
12 Giga RAM
16 Giga SAS -
Hey Juan
The monitor is calibrated with an X Rite i1.
But even if it wasn’t calibrated, Resolve would still behave like that.
Im uploading and sending you the drp,
I cant see what prefs page you want me to grab, It sounds like a mac…
I’m on a win 7 box, running Resolve Lite 10b, gtx680 card.
Thanks -
No, a wacom stylus.
I can take this to a post house that has a panel, to check this out,
thanks -
I actually work like that. I check the Arri LUT, sometimes it gives me a good result, but many times it
looks wrong, too contrasty, colors become to blue and other oddities.
On this job, I have a few clips that took the Arri LUT nicely, and many others that I correct myself.
As you say, first node is for primary and overall level and than more tweaks.
I worked like that in a few cases before and noticed Resolve’s, or my system config + Resolve, “lack of color space” behavior, and I actually thought this is because I don’t have a Panel,
but this project is too much, this is close to unusable.
Maybe it’s the version 10 beta?
But again, I did 2 jobs before this and it behaved flawlessly. -
[Itai Bachar] “What about those Eizos and NEC pa’s, or any other high end pc monitor?”
I’m thinking of getting an Eizo CX240 or CS230. They can do 10 bit if you use Display port. But you would need a Blackmagic card to send out a correct video signal. I was told that a set up using these monitors and Resolve’s Monitor output LUT for 3D LUT calibration is a valid workflow.
-
[Itai Bachar] “In what way apple monitors are worse?”
One part of my reservations about Apple generally is that they have consistently shown a fairly limited understanding of SMPTE/EBU CCIR practices. They cannot decide what gamma to use for display transfer functions and this has lead to an almost universal mistrust of exported media. Depending on decoders, displays, and operating systems, you *might*, but usually *might not* get back what you thought you were sending out. Beyond that, at least the Apple Cinema Displays are *not* designed to tell the truth — they are juiced to make everything look “great”… and they are all glossy now, you can’t get the matte low-glare glass anymore. Its actually harder to find something ‘right’ with them than it is to list everything that is wrong with them, which is just about everything. The power supplies fail regularly, and it is pretty much impossible to fix or replace them… shall I go on?
As far as the green/magenta flip/flop, its another argument against using mouse/stylus controls as a User Interface. Even back when we were exclusively grading in SD, daVinci systems were famous for over-sampling, not just at 4:4:4, but at 8:8:8 (which resulted in the name of their intermediate SD grade system). The current spec for Resolve is to process the source data in 32-bit float, which should have a very high degree of fine-ness.
It sort of reminds me of an incident that occurred while there was a pan-and-scan job in the telecine gate a number of years ago. A trainee was assisting, and noted that my X-Y coordinates were “not exactly” half of the maximum/minimum excursion values for a centred image. One issue with that is that the 8:8:8 displayed all its parameters in hexadecimal… so math in base-16 is not the same as base-10, that’s just one issue, but the other reality is that interlaced 525i29.97 doesn’t really have a 0/0 center value — in the world of 262.5 interlaced lines, 240 of which are active in any particular field, there will always be a slight geometric offset, but nothing below .5% is gong to make any difference at all because we are dealing, once again, with a sampled system. That means something else in the greater field of Communication Theory. So the reality was that there was no “line” that would be exactly centred vertically… its an even number of lines, so half would have to be above a certain point, and the other half would need to be below that value. Its a little like the joke that half of the people you might meet are “below average.”
You might note that you can really only adjust blur/mist in Resolve in 2-pixel increments (that has an effect), and certainly only even-number amounts are actually saved and retrieved. You mibght get the feeling that what we do is an inexact science…
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
perhaps i dont get it, but:
isnt the picture still greenish after the correction?and: even with the mouse you are able to finetune much subtler changes,
only the displayed numbers change “later”. -
Itai actually sent me his project. I found the same. Adjusting with the mouse while looking at the digits indicator resulted in steppy adjustments, but there was actually much finer graduation available.
One technique I suggested to him, was to have one node where he doubled the saturation to 100 then another node where he would make his adjustments, then a third node where he would half the saturation to 25. This gives the second node a much finer level of operation in terms of color balance.
That said, this is a non-issue if one doesn’t look at the numbers, or if one uses the panel interface instead and again, doesn’t look at the numbers.
https://JuanSalvo.com
https://theColourSpace.com
Reply to this Discussion! Login or Sign Up