I truly love AP, but colour limitations are now showing. Is there hope?

[nickl63167969]

Ok, so there's lots I like about AP; the editor, dynamic link, automation and much more, and so far it's been my colour tool of choice.  I can also live with the crashing (simply adjusting the scale control was crashing every time on a recent project with several nesting levels). Some limitations are starting to bite though.

As an example, a single Lumetri effect can be insufficient, but only the final Lumetri can be controlled via the dedicated Lumetri window as there's no concept of picking which one to adjust (discussed by others in Re: Lumetri: Two and more Lumetri effects on a clip ). Further, the Tangent surfaces seem only to be able to control the final effect too. Reordering the effects temporarily is one solution, but in general that's likely to cause issues so you have to start expanding properties in the effects control panel itself. Why you can't select which Lumetri you want to control bafffles me, and it slows down work if you need to adjust earlier effects. Suggestions in the thread to use SG rather than Lumetri is also a no go now that it's incompatible.

Tracking is another weakness. I'm using a decent new setup which has taken render times literally from hours to minutes over my previous one, however tracking is still painfully slow. In contrast, my tests in Resolve give realtime tracking on "power windows".

Being able to view multiple clips or selected parts of a timeline simultaneously in a bin type view would also be useful for matching between shots (a feature of Baselight).

I've glanced at Resolve and been happy to see the many ways where AP is better IMO, however there's a flipside too, and the limitations and ways from a colourists point of view that AP could be and needs to be improved are really stacking up now, which truly pains me to realise; the reality of product development being what it is, I doubt that the suggestions mechanism will pay off as well.

I've researched possible workflows to roundtrip Resolve and AP, but while possible, realistically it doesn't seem the way to go.

So I suppose my question is whether there's a way to beef up AP from a colour and perhaps tracking point of view (in a way that's better than deferring to Mocha) as I'd like to buck the trend of using Resolve and stay with AP for all its strong points. (I will be taking a look at Colorista.)

[R Neil Haugen]

You're right on about the issue with multiple Lumetri instances on a clip ... and it's a real pain in the patookie. NOT advised to move the instances around, however ... PrPro is very linear in the way it applies the effects/controls: as you go down the Effect Control Panel, the effects are applied. And as you go down the Lumetri panel, the effects are applied ... in order top to bottom.

If you've got a Secondary in a second instance, but move it "up" to work the previous with your  controls ... that Secondary is now applied before what was the first instance. So when you're working the (now) lower panel, you're working after the key. Move them back, you're probably going to wonder what happened with your pretty pixels as the key is applied in a different place.

There's quite a number of other issues ... from more of a colorist's perspective. There's two LUT/Look application slots, one for camera/tech LUTs at the top of the Basic tab so it's the very first thing applied to the clip, the other in the second tab, Creative, where supposedly you would put print-stock type LUTs to use.

Problem is, no one on the planet suggests those locations for such LUTs. Tech/camera LUTs are supposed to go after your basic neutralization/white-balance controls so you feed a corrected signal to the LUT. And print-stock LUTs ... well, if you want a final product that's a 70's soft shadows/highlights movie look, and you drop it on ... then do ANYTHING in the Creative, Curves, or Color Wheels tabs ... you've now screwed up your carefully chosen look. You can only use that slot to put something you're expecting to modify out of whatever it once was. Which is apparently all the designers of this thought "we" would want. Really. Sigh.

Some things you can get around by using Adjustment Layers over your clip ... if you've got one Lumetri applied per AL, you can of course use the Lumetri panel ... and your Tangent gear ... with each layer. From my testing, the application there is sort of upside-down kind of ... I think the stuff applied to the clip is processed first, the next layer up next, the next layer up next ... I need to do more testing to be absolutely certain of that but so far that's worked for me.

The one handy thing about the placement of the "tech" LUT slot at the top of the basic tab, is if your media is seen by Lumetri as having values below 1 or above 99 on the left side of the RGB Parade/Waveform scopes, you can drop a pre-built LUT/Look in there that rolls off the media so the signal falls within 1-99, and then be able to properly use the Basic tab. As they have set it so the Basic tab controls are NOT allowed to touch data below 1/above 99, you get very weird results if you try to move that data within "bounds" just using Basic tab controls. The out-of-bounds data stays put, "super-blacks" and "over-brights" both, but you can move the data within bounds around. Making horrid banding of course.

So I've built a couple variant LUTs to drop there, named so they are the first couple things to get to via the drop-down/right-arrow routine. Otherwise, the first Lumetri panel that can successfully "touch" the media is the Curves tab .... Color Wheels are also allowed to work on that out-of-bounds signal.

For my work, one of the biggest issues is no shot-matching capabilities ... or even procedures ... built in. Which is why I use Patrick's little app here to modify some of my 2017 projects so I can still use them in "Direct Link" mode with SpeedGrade 2015.1. (Referenced in this thread on the SpeedGrade forum: Update: My little Speedgrade/Premiere Project Converter still works with 2017 .)

You then run your PrPro project back through his little app, and go back to working the project in PrPro 2017. It's a HECK of a lot easier and faster than exporting an XML and flattened timeline to Resolve, conforming there, and round-tripping back to PrPro!

I do have a quasi-shotmatching setup I've created in PrPro ... I've got a keyboard short to turn off Mercury Playback on a second screen (normally on for all other work), and a custom "Shot-matching" workspace I created and put on the Workspace bar. It's got a second panel that is placed to cover the monitor I use for playback work, with the Source, Reference, and Program monitors on it in three sections left to right. I only use the Source for stills reference shots, so normally it's only the far left inch of the screen.

The Reference and Program monitors take up the majority of the screen, and I use the image size/percentage controls and the scrolling tools to get the parts of each I need at the size that works best, use the timeline of the Reference monitor to get to a good frame of the clip I'm matching to, and the main timeline to set the program monitor to the clip/frame I am going to correct. Clicking forth & back between images can get either on the Lumetri scopes which are always out in this worskpace, which is a big help.

It ... works ... kind of ... but there's not an auto-shot-match button as in SpeedGrade, which typically gets me 80-90% of the way there with a click ... nor is this any real replacement for a decent 2-up/3-up monitor setup. Which is why I quite frequently go through Patrick's app to SpeedGrade. AfterEffects comps are a problem for that, so if you've got those, you need to export a full DI clip out of  AE, and replace the comp on the sequence with that AE exported clip.

I have found that by totally mapping out the Basic tab controls for both the Ripple and the Elements panels, that is suddenly a pretty decent "neutralization" tab. I've got the Blacks, Exposure, and Whites controls mapped to the Ripple Dials and outer rings of the Elements; and left-to right on the balls, Contrast (horizontal axis) versus Saturation (vertical axis); Temperature (hor. axis) versus Tint (ver. axis); and Shadows (hor. axis) versus Highlights (ver. axis).

So ... small project, a few clips, nothing much ... I'll stay in PrPro, use the Ripple or Elements and do neutralization in Basic, Creative, Curves, or more typically Color Wheels to touch the clip for a better look or match, secondary at need, and be done with it. If I do need to more closely match a few clips, I can go to the shot-matching workspace and do ok. Then apply an AL for a total 'look' over a scene or project.

Much of a project ... edit in PrPro, "open" the project file in Patrick's app, take the output of that into SpeedGrade for color, save and "open" the project file back in Patrick's app, and it's ready for work back in PrPro 2017.

I would LOVE it if you'd file a ton of bug/feature forms about the parts of Lumetri that need work ... and if you're interested, returning Direct Link to SpeedGrade ...

https://www.adobe.com/cfusion/mmform/index.cfm?name=wishform

Oh ... tracking: yeah, PrPro's tracking isn't particularly totally spiffy, that's another b/f report form ...

Neil

[nickl63167969]

Thanks Neil. You're right, and though it pains me to say it, I'm converted to Resolve. I started to look at it in detail last night and the kinds of issues inherent with AP that bugged me aren't there. Even when Resolve crashed (I hit bugs in the Nodes editor), its crash catcher allows you to carry on without it bailing out. Rotoscoping isn't as powerful as AE, but tracking is so much faster and accurate, the time gained in faster tracking you can spend on tweaking beziers and still be ahead overall.  Being able to view and scrub through multiple clips side by side in parallel that I'd seen in Baselight is useful, and the variety of different node types such as layer vs parallel and easy node mutation makes setting up corrections trivial in comparison. A scene auto detect feature works pretty well, setting up optimised media for clips (i.e. image sequences) is fast and again trivial in comparison. I prefer AP's more intuitive naming of features, such as blacks vs "Lift", but that's just wording and not any functional advantage. I like Adobe's history mechanism, being able to jump back and forth over multiple changes, but can achieve similar in Resolve. Versions (on my list to explore shortly) is a welcome feature, having grouped clips you can apply pre and post group nodes for common grading, and trivially apply one common grading say to clips 1,3,5,7 in the timeline and a different one to clips 2,4,6,8.  Doing an EDL based round trip from AP into Resolve (not gone back yet) isn't ideal but worked on tests I tried. XML/AAF not so pleasant. AP has the edge in some UI aspects such as combining shift with left/right in timelines for faster movement and being able to click/drag the timecode that I believe Walter Murch asked them to do, but the UI overall isn't too bad. I did try Colorista, but that's really just a kludge and it doesn't overcome the underlying issues. I've more essential workflow processes to cover, but so far it's been a breeze to migrate.  Did you ever try Resolve yourself in any detail?

[excited_Genie16B8]

having grouped clips you can apply pre and post group nodes for common grading, and trivially apply one common grading say to clips 1,3,5,7 in the timeline and a different one to clips 2,4,6,8.

Interesting.  There's an Adjustment Layer feature request going around for Resolve 13 because it's believed what you claim here just isn't possible without them.

I'm only 200 pages into a 1200 page manual, though.  Still have LOTS to learn.

[nickl63167969]

Groups are explained here DaVinci Resolve 12 - 66 Using Groups - YouTube

For anyone who knows what they want to find and/or what should be in the software and they want to know where to find it, Daria's "let's get to it" approach is great.

Groups let you group clips together as you'd expect, and then in the nodes editor there are two other options available on the clip dropdown, Group pre-clip and Group post-clip. The nodes in each of those get applied in the order suggested, with nodes for the individual clips applied in the middle. I'd see an adjustment layer as useful to apply an overall grade to absolutely everything, going further than using remote grades, if that's not possible already. Remote grades seem similar to what you would achieve in AP with master effects in that they apply to the entire source footage. Hope I've got all that right.

[R Neil Haugen]

I've got Resolve up, and have done some small work in it. I'm also a member of the Mixing Light subscription-service of colorists Patrick Inhofer, Robbie Carman, and Dan Moran. Robbie and Patrick are both in a LOT of lynda-dot-com tutorial series involving SpeedGrade or Resolve or even Premiere Pro. Also at least Robbie is in some of the AdobeTV tutorials, and I think Patrick is featured in a couple tut's on the Resolve site. I met them all at NAB2014/Vegas, and have had the pleasure & honor of sitting to table with them at several events, and assisting on last year's Colorist Mixer sponsored by Team Mixing Light. Great guys, SO knowledgeable ... and so helpful. I TA'd several of Robbie's classes at Adobe-MAX/San Diego last November also.

They have many tutorials on their main site on the use of Resolve ... Patrick is about done with what will be at least a 9-part series on conforming a project edited in PrPro into Resolve. For colorists, they really prefer XML's to EDLs, but everyone's mileage always varies. How you name & then "ingest" your files into PrPro apparently can make a big difference in how well say EDL's out work in either SpeedGrade or Resolve. For a slightly higher fee (over the base subscription) they've got an amazing several-hour program of detailed Resolve training, including downloads so you can work the same material. I've taken a "taste" section of that, and it's by far the best Resolve training I've seen anywhere. Yea ... that good!

Weekly Color Correction Tutorials and Articles Library

Really, to use Resolve the way I'd prefer on my twin-screen setup, I need to get the $100 BM external box that Resolve requires to run a monitor as a "Program" monitor. I've not really enjoyed the UI in two screens with a small playback monitor as my only option. Just haven't pulled the trigger yet.

Jim_Simon​ 's comment about being only 200 pages into a 1200 page manual ... YES! THEY HAVE A REAL HONEST-TO-GOSH USEFUL AND DETAILED MANUAL! And yes, I'M SHOUTING!

Wish that Adobe would provide such. The Powers That Be have 'heard' my comments on that, and they don't seem to need to have me to lunch to chat about it ... funny that. Ah well, I'll survive the slight. But Jim, aint' it grand to have that available?

So, for my small projects, really learning Resolve just hasn't been necessary ... with SpeedGrade, I can do what I need that Lumetri can't. And for some of my stuff, realistically, using the Ripple or Elements panels in Lumetri works great, plenty fast enough, and I'm happy. The use of external controls is the ONE thing that makes Lumetri really useful, to me. Without the externals, to me that Basic tab is a waste of time. Mapped out to my control surfaces, it's actually a pretty good neutralization panel.

And Patrick Zadrobilek's little app is very fast in modifying a 2017 PrPro project file to use in Direct Link mode opened in SpeedGrade 2015.1 ... and then to return the project file to 2017 status after finishing in SpeedGrade. For me it's adequate and far faster than an EDL/XML out to Resolve, then round-tripping back.

For a one-man shop, the whole round-trip thing can become manageable in a fairly quick process, you learn what you have to do rather painfully by when you don't nail all the steps right. Talking with "real" colorists who do work from all over, that whole conforming thing takes normally at least one hour of in-coming time per project, and can easily become three to eight hours depending on the size of the project, care of the editors in sending out assets, amount of time-ramped material, and about 3,000 other factors.

Now ... what I have played with in Resolve, yea, that's an app with incredible controls. Holy Jeepers Batman! Of course, learning how to use 20% of those controls well ... is a HUGE job. Still, what you can do with a few hours of play-time and a few tut's is wondrous.

And of course, that program really allows something like the Elements panel to shine. Robbie & Pat say that with a mouse, they can grade (to their pro level of need) 100 shots or so a day. The Ripple takes them up to a bit over 250. The Elements to 500 or so. And Robbie says he can touch up to 1,000 with the full Resolve panel.

And Resolve is built sensibly ... any LUT you apply to a node is applied to the signal after any other things you change in that node. So you can go ahead and apply a RED or Arri camera-corrective tech LUT to your first node, then within that node, adjust the controls to neutralize the clip ... and you're doing it by feeding the LUT the proper signal, the way the camera makers expect.

In Lumetri, it's bass-ackwards. The correction on the Basic tab happens after the tech LUT is applied to the media. So ... I use the LUT slot in the Basic tab to say apply a corrective LUT to the media to bring highlights down within the 1-99 scale limited area the Basic tab can touch ... and any tech-corrective LUT goes in the LUT slot on the Creative tab. THEN I work the Basic tab to neutralize. It works ... but Resolve is just built correctly.

We could all three go on all day about where that's built correctly though ... which is the sad point about this thread, isn't it?

Now ... if the management for the PrPro team could re-think a few things ... wouldn't that be marvy?

Neil

[excited_Genie16B8]

That is very nice indeed and would seemingly handle the need for an Adjustment Layer.

[nickl63167969]

Right, for many cases I would say. I looked at BMD's thread on feature requests after you mentioned AL's as I wanted to read other's use cases, and saw a comment "if Blackmagic truly has aspirations for Resolve to become a full-flegded NLE, competing with the likes of Premier ". So a similar thread to this might be had over there from the perspective of editing features. It's definitely not all rosy with Resolve though. After posting when experimenting more I got Resolve into a case where it would always fatally crash when playing towards the end of clips that weren't in a group, and posted a report in the forum with screenshot to give context, and that never appeared and seemed to go into a black hole; maybe it's still pending moderation but I wonder if they delete bug reports. That's a bit worrisome.

As for other developers and manufacturers, the lack of an API is a shame too as I'd like to knock up a custom control surface; hardware wise it's easy enough, and when I scoped out the idea it's fairly straightforward to drive parts of AP, but a no go for Resolve unless using a mouse/keyboard emulator and adjusting a single control at a time.

I'm glad I started with AP though so I can choose between either as necessary. Good luck with your own Resolve learning!

[R Neil Haugen]

The Ripple by Tangent does work for some things within Resolve, and of course, does very well with PrPro/SpeedGrade, and is about $350 USD. Not as full featured as the Elements panel, of course ... but still useful.

Neil

[nickl63167969]

I have a Ripple connected but overall I prefer working with an M570 trackball and find I can make initial adjustments quicker with that. For making finer changes the Ripple seems to work ok though.

[R Neil Haugen]

Getting the Ripple set for personal preference for "granularity" of change is a small chore. You can set it so that you have a more coarse level and a finer level, that you swap via one of the two buttons.

Neil

[nickl63167969]

Ah, that's interesting, thanks Neil.

It's not taken long, but I have to say that the initial good vibes with Resolve have worn off rather. I heard an audio glitch in some music after making a compound clip (a nest in AP), and found that it dropped around 50ms of audio at the end of the compound clip. After opening the clip, Resolve also crashed fatally when trying to export its audio. That's the second fatal crash bug hit in as many days doing trivial things. The original and what I saw in the waveform after rendering are below. Never had these type of issues in AP.

[excited_Genie16B8]

a similar thread to this might be had [in the Resolve forum] from the perspective of editing features.

Most definitely.  Where PP lacks in color, Resolve lacks in editing (and delivery).

[excited_Genie16B8]

the lack of an API is a shame too as I'd like to knock up a custom control surface

Well, there you'd be cutting into their $30,000 Control Surface sales.  I can understand why BMD doesn't want you doing it on your own.

[R Neil Haugen]

Robbie Carman and .Pat Inhofer note that with a mouse, then can maybe do 100 or a few more shots a day (colorist) and do their top work. Elements panels a bit more than double that, say 250. The full Resolve surface? At least 500, and I know Robbie says some days he can make close to 1,000 clips with the big surface.

At that point, and for the billings per hour they can get much of the time, that big panel makes financial sense.

For most of us, well ... not so much, right?

I ain't never gonna be the target market for the Surface ...

Neil

[excited_Genie16B8]

I ain't never gonna be the target market for the Surface ...

You wouldn't want it if Nick built you one for $100?

[R Neil Haugen]

Well now, that's the horse of a different color, right?

Neil

[excited_Genie16B8]

There's my point.  The essence of Trademark, Copyright, proprietary technology, etc is "Pay us, or do without."  Somebody comes along and offers another option kind of invalidates the whole system.

[nickl63167969]

Realistically you'd need to add a zero at the end but it would solve the craziness required just to make a minor parameter change with a Tangent element.

panel rings - YouTube