Joseph Owens
16
COW Coins
16
DaVinci Coins
Forum Replies Created
-
If you can keep saving stills so that there is some capability for grading in some kind of context, with this wild mixture of formats, you might be better served by separating some of them. Frame rate differences are a big enough hassle — start mixing in the probable differences in aspect ratio is going to start building an unrewarding day, trying to deal with all of them at once. If you are unsure at this point, it is just going to get worse.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
[John Riker] “I have a bunch of VHS converted content I’m trying to improve the coloring on “
Does sound as if Resolve may not be the correct choice for you.
There are a number of plugin processes available across the net — Boris, Sapphire, etc., that may get you to where you want to go without the hardware demands and learning curve that a fundamental tool like Resolve demands.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
[Paul Willis] “I have been dubious of software legalisers”
For the most part they are simulators trying to predict what will happen (based on RS170A composite baseband values) within a 0-100 RGB math model. As Juan responded, the real environment is the combination of luminance and chrominance (Y+C) which requires an encoding matrix.
In general, you are allowed -2 to 102 IRE in Luminance (on a 601 digital scale with a 0-100 luminance band, excluding analog pedestal of 7.5 IRE) and -20 to +120 Y+C for chrominance, which you cannot directly measure in a strictly RGB system.
The software systems say that you are okay if none of the RGB values are outside the 0-100 gamut — in which case for example, color bars would be illegal, if they included the I and Q flags as those test patches generate illegal (negative) RGB values, even though they are part of the NTSC system.
Not trivial.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
[Sascha Engel] “Any idea why”
First guess would be that the multicamera sequence is being handled in a similar fashion to the import strategy that would be applied to either a completed export movie that would be marked out with either SceneDetect or an EDL. Even though the master is a single clip, the way the events are represented in the Resolve timeline are as independent clips. There is no way of “joining through” any of the clips in a SceneDetect import, either.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
In a sense, this is how the Linux version distributes processing.
Remember “Apple QMaster?” It didn’t work, either.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
[Tim Franklin] “how do you use only a specified channel from one layer to grade the main image in Davinci?”
This varies slightly from one version to another… since in later versions of Resolve you can split out RGBA channels discretely to pass along to other processing nodes.
In its simplest approach, if there is a discrete travelling matte movie, which can be a straight black&white QT and doesn’t need to be RGBA, there is a specific import procedure that associates it as a matte, mated with a primary source clip. Essentially, you select the foreground or finished composite plate, then import the alpha movie as a matte key element. In the COLOR page, add a second node to the grade for the primary composite source clip, right click on that second node and there should be a dialog allowing you to select the associated matte movie as a qualifying element. If you need to do inside and outside corrections, then you also need to add an “outside” node. The manual goes into some detail outlining how to implement matte clips or primary source clips that have an imbedded alpha channel. If you are actually layering plates, it is possible to add an alpha output to a node tree that creates transparency between video layers on the Resolve timeline.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
[bea vilarino] “exporting all the footage to DPX “
If that works, the problem is with Quicktime, not Resolve.
If you widened your search criteria for other threads concerning quicktime differences with rendered values (you did do a search, correct?) you’d discover several thousand discussions over the past 8 or 9 years struggling with problems with value scaling, gamma, and many other mis-matches.
Without any details of how you are actually judging the differences between your expected and observed results, its really impossible to point to anything other than the usual explanations — looking at your examples, though, it suggests that the only saturated color in the image is in fact the flesh tones… if there were a larger range of hues, my bet is that you would more than likely see the entire gamut being de-saturated. My other guess is that you likely haven’t measured any of these possible differences with a waveform monitor for an objective evaluation. Generally that makes a more accurate evaluation fairly obvious, and can point to where the results have diverged from your intended goal.
If you are depending on Apple OS to supply accurate results, sadly, that is overestimating that company’s ability to deliver consistent output.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
[jake blackstone] ” fix the bug preventing the joining of the wrong cuts in the timeline after the scene detects!”
Not a bug per se, because that would infer there was some process that would allow for it, were it simply an option. It is a workflow issue — scene detect is part of the “import” process, so the application simulates the creation of independent clips, which… cannot be “joined.”
That’s what’s really wrong with the process, and its been more or less the same engine that daVinci has been using since the Renaissance. No, not that Renaissance, the one that came after the Leo.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
[Martin Halm] ” i also tried to deliver jpg2000 straight out of resolve. open_dcp won’t accept these files. strange ..? “
From what I understand of OpenDCP, it wants to do the J2K conversion (and XYZ conversion) internally and make its own MXF folder, and for that it is going to want TIFF or DPX (essentially you would be making a DCDI first).
Most DCP authoring packages don’t really lend themselves to accepting a bunch of files from different directions, as the CPL/PKL/Assetmap XMLs have internal UUID parameters that must integrate OR ELSE.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.
-
Can you MarkIn/MarkOut those special cases and try rendering a target clip for each?
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.