[Marc Wielage]

[alex gat] “I was looking at how to combine timelines in davinci and I dug up this old thread. It is seems that even today it is not possible to combine timelines in davinci. But then how do you guys edit long feature films ?”
Starting with v12, there will be a way to nest multiple timelines and output one big file of an entire feature.

Otherwise, as Michael says above, you can cut and paste entire timelines if you’re careful.

[Marc Wielage]

In worst-case scenarios (particularly with missing reel numbers or timecode issues), if you load the camera files first and then load the XMLs, Resolve will find the material and will drop it correctly into the timecode. On the rare cases where the files aren’t found, you can always manually force the file to conform.

You can also try Import -> Additional clips with Loose Filename Match. In some cases, you may find that the AAF results in better results than an XML (or vice-versa). I also generally have the client give me an EDL, and even that can work if you have a very simple timeline. At the worst, it will help you locate where missing clips are.

I’m not a fan of using H.264 material for editing or color-correction in general, and my preference is to transcode this material to ProRes (or DNxHD) first, because I think it works better with Resolve.

[Marc Wielage]

Upload it to Vimeo and see if that looks any better. These gradiated shadows are a worst-case scenario for compression, because it’ll tend to exaggerate the 8-bit stepping. Not enough bits.

There are tutorials out there on the web that will explain how to optimize your material for upload to YouTube and Vimeo with the fewest possible artifacts. I generally work in ProRes 444 and post a ProRes 422 to Vimeo and let them handle the conversion within the site, and the results are acceptable to me when starting out with 10-bit material.

[Marc Wielage]

I totally agree with Joe. I don’t always split up functions across multiple nodes, but I do sometimes just in the event the client says “I like part of this correction but not the other part.” That way, you can delete one node and give them what they want.

Node organization is a hugely important topic. I tend to do a “normalizing” correction in Node 1 (including camera raw settings if necessary), basic color in Node 2, fine-tuning in 3, followed by windows, keys, curves, and defocus (not always in that order). I typically will do NR last, but it depends on how much I’m leaning on caching. Multiple power windows could easily require 5 or 6 nodes, possibly more.

I have seen colorists that layer 25 or 30 nodes per shot, but I usually have to work fast enough that I’m lucky to hit 9 or 10. For features & episodic TV, I think figuring out how many minutes per shot limits the number of nodes you can practically do in the available time. Commercials are a completely different deal, and if you can spend 4 hours on a :30 second spot, you can certainly spend 15-20 minutes per shot with no problem.

I know reality TV colorists who have to blow through 700-800 shots in a 6-hour session, so they’re lucky to do 2 nodes per shot. All you can do is solve basic matching issues and let the subtleties go.

Keeping the node structure identical on every project helps people work faster, since you can figure out where and when to use PowerGrades, drop in Pre/Post-Clip grades, and so on. All the usual tutorials out there deal with this; I think Patrick Inhofer on Mixing Light has devoted several Insights on the subject, and Alexis Van Hurkman has talked about his observations on consistent node structure.

[Marc Wielage]

I’m seeing a weird double image (chroma only) on the far right side. I’m assuming that shouldn’t be there.

I wonder if something is not being converted correctly in the transcode. What format are you shooting in? What format are you editing and color-correcting in?

The last 5D project I did, I had them transcode everything to ProRes 422HQ (and the HQ really wasn’t necessary, but I figured what the hell), and it came out surprisingly well, even on a 20′ screen. I think one key going in is knowing that the material is only 8-bit and you can only push it so hard. I kept the blacks on the somewhat-crushed side and also used a lot of vignettes in certain scenes.

The grayscale macroblocking is one of those 8-bit things you can’t overcome. As a last gasp, you could try to overlay some film grain on top of it and see if artificial dither reduces the stepping. These 8-bit cameras are really awful to deal with in situations like this.

BTW, you should definitely fix the audio — the singer is way, way off mic and too echoey (in my opinion).

[Marc Wielage]

You can add a BMD video card (like a MiniMonitor) and see the main Viewer display on a third monitor. That’s the only way to do what you’re asking, as far as I know. Many, many editors work this way with Avid, Premiere, and FCP.

[Marc Wielage]

Shoot color charts on both under the same lighting conditions, and see how they look.

All of this can be done with a calibrated monitor, scopes, experience, and time. Be warned that sometimes you have to take the better camera and degrade it to look like the cheaper camera. It’s rarely possible in my experience to take a cheap camera and improve it to make it look like a more expensive camera. I have done shows that had Red, mid-level Sony, and Canon 5Ds shooting the same scene, and much of the time the best I could do was whack the Red and Sony footage and make it look as bad as the 5D. It all matched and the director was happy with the results.

[Marc Wielage]

[Neil Sadwelkar] “One more thing. In a timeline of ArriRaw and Sony F65 mixed, I had to move the F65 files to another layer and then do consolidate on ArriRaw and F65 in separate passes, else it doesn’t work for the F65 – get full clips.”
I recently had a similar situation with CinemaDNG clips and R3D clips, and came up with the exact same solution. At times like this, I really wish that crews would shoot on just one kind of camera… but sadly, I think those days are gone.

[Marc Wielage]

[Rick van den Berg] “i use it in my office with a flanders monitor and it works great. but i cant believe there isnt another way. every other piece of software like premiere and final cut can do it.”
Every color correction program out there also is typically a 3-monitor setup: one for the GUI, one for scopes, and one for the color monitor. I know of many editors who cut with 3 displays: one for bins and other functions; one for the timeline and source/timeline viewer, and one as a larger hero display that just shows the video output. For me, it’d be very hard to cut without three monitors. Color-correction is almost impossible, at least with the way most colorists I know work every day.

[Marc Wielage]

Sounds to me like the OFX plug-in is getting delayed when the frame boundary is hit. I would contact Film Convert and yell at them — this may not be a Resolve issue per se. There might be a way for them to provide an overall setting that could dial in the “frame early/frame late” number to get it correct.

I have seen other OFX plug-ins behave badly with keyframes as well, forcing me to start them earlier or later than expected in order to get the desired results. I haven’t seen them totally miss a frame… yet.