Red to Film Output

[jake blackstone]

Yes, it was a typo, obviously I meant to say PD985...

 

[jake blackstone]

That is not correct. What they recommend is PDLog985 with a 2 stop pull. The problem with PD685 is that the available levels are not used optimally as all values are forced to be between 95 and 685, even if those levels should actually be "super black" or "super white." By using PD985, the values are spread more over the available range, as they should be.

Redlog is another option, and a good one if you know how to handle it properly. Of course, if you happen to be using Scratch, there are other options that are even simpler and more accurate. And none of these choices make any sense unless you a) understand what you're dealing with, and b) have a proper film print preview LUT in place in the first place. If you don't, Andrew's advice makes a lot more sense than trying to guess as to what it's going to ultimately look like.

I'm just curious as to what other options, that are "simpler and more accurate", that can be employed in terms of color space and gamma management in Scratch vs. any other basic transcoding solutions, such as RR, RC, RA etc...

 

[M Most]

I'm just curious as to what other options, that are "simpler and more accurate", that can be employed in terms of color space and gamma management in Scratch vs. any other basic transcoding solutions, such as RR, RC, RA etc...

How about one setting for "Lin" and one setting for "Log" that work directly from the RAW data, automatically remove the camera metadata, and only require a user supplied gamma setting? And are globally applied?

(Obviously, I'm not necessarily talking about dailies here, I'm talking more about finishing..)

 

[jake blackstone]

How about one setting for "Lin" and one setting for "Log" that work directly from the RAW data, automatically remove the camera metadata, and only require a user supplied gamma setting? And are globally applied?

(Obviously, I'm not necessarily talking about dailies here, I'm talking more about finishing..)

And how is that more accurate and or faster? There can never be one setting for grading and this is obvious. Every shot in the timeline needs different settings (exposure, color temperature etc). That is why I do not believe in one LIN and one LOG setting, that it can be somehow magically applied on the global basis. Beside, any decent grading app (Baselight, Lustre, Pablo, DaVinci, Filmaster etc) can easily apply the same or just user selected settings to the rest of the timeline with one click. Once it's applied, it overrides the metadata. Apparently you're talking about the grading aspect of Scratch and not just color space and gamma "applicator", therefore we need to start compare apples to apples. There is nothing fundamentally different about Scratch approach vs. any other grading solutions. Granted, my original question was about color space and gamma settings, that can be applied to RAW files during the transcode to LOG DPX in PD985 for finishing grade in Luster (as was recommended in the Autodesk White paper). Your answer implied, that somehow Scratch is "faster and more accurate" than anything else out there in that respect. Although Scratch is an excellent conforming and a decent grading application, it has a long way to go, before it will catch up to the top grading solutions. And that is why I always finish my grade in Lustre or Baselight

 

[jake blackstone]

Andrew, Duart, LaserPasific, Efilms, Technicolor and others are a point of reference for all of us out there, outside of Hollywood... so when we hear a suggestion we put a lot weight on that opinion... we start testing it, same of us we depend whole projects on them... a more in depth with the pros and cons would be better, people want always the best for the less money possible... its a give and take process... Log is much better but not at indie budged, indies they need to know that.

Karilla, please explain your experience so far with REC709 workflows, with two different post houses that using few of the best REC709 monitor available... in comparison to what we do together in log... without even know me in person, just a voice over the phone, the first movie is in Cannes with just the zero print and NO further corrections and now its the second feature that we do together...

The quote from the producer of the first film that we did with Ido:

"Dear Evangelos,

Many many thanks for a job extremely well done and for the wonderful hospitality by you and your charming brother.
Looking forward to working with you again in the near future.
Next time I will not come in Athens, I didn't believe it, now I know, I don't need a corrected copy, zero copy is immediately perfect.
I will be screening the print to a bunch of the top Israeli film producers and I will gladly recommend working with you.

Best,

Haim"


So to wrap up...

• REDLog or PD985 with -1 to -2 in exposure is the best source to start with.
  
• Until now, Redcine delivers better debayer than Color.
  
• Log workflow has a film response, that means, it doesn't clip highlights nor blacks, it just compress them. REC709 or linear is just clipping...
  
• If there is absolute NO chance a project to reach film, then REC709 is possible...
  
• Even if you are NOT going to film, grading with a Film print LUT in Log always deliver, more Cinematic images, that then, with a LOG to REC709 conversion LUT can deliver the best results for DCI, DVD, Broadcast and BlueRay
  
• The linear workflow VS Log workflow, for Filmout there is no contest... Log is much better in many ways, RED could be linear, but film is LOG and if you want to play in the film playground you have to play by the rules... film rules... film understands only Log...
  
• You can do a REC709 workflow, but don't expect THE BEST results...
  
• If you want a Video look, DON'T do a Log workflow...
  
• Log workflow for filmout needs a proper color managed LUT driven workflow and that is a fact.-

And as Mikes points out "And none of these choices make any sense unless you understand what you're dealing with"

Just to expand a little on the modern color management. In the old days of film mastering to DVD we had to do multiple passes for Pan and scan 4:3 PAL, 4:3 NTSC, 16:9 PAL, 16:9 NTSC etc. Additionally, Rank Cintel would drift so much, that we would have to do the global grade touch up, just before the laydown to tape. Luckily, everything then was either SMPTE 601 or later REC 709, easy... Fast forward to today. One LOG 2K DI for everything. DCI, filmout, HD, NTSC and PAL. Just use the proper LUT for all of the conversions. Not so easy, but much cheaper and faster...

 

[M Most]

And that is why I always finish my grade in Lustre or Baselight

Well, congratulations. I'm sure that makes an even bigger difference than having a talented colorist.

Seriously - Every shot does not need different basic settings if you're working directly from the original RAW information. All of the settings you're talking about are purely metadata. They are set on the camera during production, but that doesn't mean they're optimal for the final result. That's why most professional colorists discard the metadata as the first step in doing a final grade. If you're not converting to a 10 bit format as the first step in finishing - as you have had to do up until recently if you were working with anything other than Scratch - there is no reason to apply any of the settings you're talking about for any reason other than convenience. If you really think you are getting the same base information by converting to DPX using the settings as you're describing them vs. working from the RAW data, all I can say is you've probably never worked directly from the RAW data. I have. It's not the same.

As for the specific color correction program involved, while one may have more convenience features than another, they all do basically the same thing, regardless of how it might appear to someone who likely doesn't use these things for a living. A good colorist will do a good job regardless of the system in front of him or her. And I say that as someone who's used quite a few of them. If you really believe you get a "better" result because you're on a Baselight or Lustre than you do on Scratch - or Resolve, or even Apple Color - then nothing I say will convince you otherwise. So I'm not going to try.

 

[M Most]

Just to expand a little on the modern color management. In the old days of film mastering to DVD we had to do multiple passes for Pan and scan 4:3 PAL, 4:3 NTSC, 16:9 PAL, 16:9 NTSC etc. Additionally, Rank Cintel would drift so much, that we would have to do the global grade touch up, just before the laydown to tape. Luckily, everything then was either SMPTE 601 or later REC 709, easy... Fast forward to today. One LOG 2K DI for everything. DCI, filmout, HD, NTSC and PAL. Just use the proper LUT for all of the conversions. Not so easy, but much cheaper and faster...

Well, in reading this it seems you've been doing this for some time, so if I implied otherwise in my previous post, I apologize. You never know what level of experience anyone on the Internet has these days. Some of the points I was trying to make still stand, though...

BTW, I go back to days long before mastering for DVD. I was around when we were mastering for VHS, Videodiscs, and RCA SelectaVision discs. And the common mastering formats were not digital, they were 2 inch IVC9000, 1 inch C, or 1 inch B formats. So in that context, you had it easy

 

[jake blackstone]

Well, in reading this it seems you've been doing this for some time, so if I implied otherwise in my previous post, I apologize. You never know what level of experience anyone on the Internet has these days. Some of the points I was trying to make still stand, though...

BTW, I go back to days long before mastering for DVD. I was around when we were mastering for VHS, Videodiscs, and RCA SelectaVision discs. And the common mastering formats were not digital, they were 2 inch IVC9000, 1 inch C, or 1 inch B formats. So in that context, you had it easy

Apologies accepted:beer Nevertheless, I still would like to reiterate my point, that Scratch has a way to go and it's not a more precise color solution, than other systems. I will try explain what I mean. I have tried using Scratch on several different occasions for grading and I just couldn't bring myself up to wade through totally incomprehensible user's guide. I hate to repeat myself, but if anyone would read any other manufacturer user's guide, they would know exactly what I mean. They usually are concise, comprehensive, informative, alphabetized and, most important, they're not written by an engineer. Bad documentation speaks volumes to me. Second, when I have a client in the room and they're paying top dollar, I can't afford to fumble for the simplest tasks, such as a key (Luster's keyer is top notch, straight out of Flame). Actually, now there are two of them. Shape tracker, multipoint tracker, cloud tracker, again the same source- Flame. And now, the most important tool for the colorist - Group Grading. Nope... Can I hide the unused shots in the group I'm grading, so I could see the flow of my shots, like I can do in Luster? Nope again. I can group my shots into groups and all of a sudden, instead of 300 grades, I'm doing 30! Now, let's talk about being able to apply 12 layers of primary grades, HSL and RGB curves, keys with garbage masks and just simple geometries on top (while tracking), bunch of effects, grain reduction, sharpening etc? Can you see it all playing back in real time? The answer is yes for Lustre and no for Scratch. And I'm just getting started Anyway, you get my point.
As far as me having it easy, the first post house I'd worked, in Dallas, we'd used two 2" Ampex VTR-1200 (not sure about the name) for editing (yes, we did use the razor and the developing solution for edits occasionally as well) with another RCA TR-600 thrown in for good measure for back up. For color camera we had used couple of Norelco PC70? I can't remember that name either, I guess I'm really dating myself

 

[M Most]

I have tried using Scratch on several different occasions for grading and I just couldn't bring myself up to wade through totally incomprehensible user's guide.

I agree with you about the questionable documentation. But documentation in itself does not impact the abilities of the product. If that were the case, Sony would have been out of the professional end of the business years ago

Shape tracker, multipoint tracker, cloud tracker, again the same source- Flame.

While this is all true, these all fall under the category of "convenience features." Do they allow you to do more in less time? Yes. Personally, I don't use keyers very much in grading because they inevitably create either noise or edge problems - I tend to use either tracked shapes, curves, or vectors. However, I know that a lot of colorists do use them, so, yes, a better and simpler keyer is an advantage.

And now, the most important tool for the colorist - Group Grading. Nope...

Well, maybe for you. It's never been particularly important for me. However.... You actually can do this in Scratch, by creating a custom Tray with the shots you want to group. Any trim you do on any one of them can be applied instantly to all of them. Same function, different implementation. And, yes, not as convenient or obvious as on Lustre, but available nonetheless.

Can I hide the unused shots in the group I'm grading, so I could see the flow of my shots, like I can do in Luster? Nope again. I can group my shots into groups and all of a sudden, instead of 300 grades, I'm doing 30!

True. Also true - this is another convenience feature, not something that provides more precise color correction.

Now, let's talk about being able to apply 12 layers of primary grades, HSL and RGB curves, keys with garbage masks and just simple geometries on top (while tracking), bunch of effects, grain reduction, sharpening etc? Can you see it all playing back in real time? The answer is yes for Lustre and no for Scratch.

Maybe on YOUR Lustre. I can show you a number of Lustre systems that it would not be true, at least not without a number of background rendering nodes.

I am certainly not arguing that Lustre (or Baselight, or Resolve) is not an excellent product. It is. But everything comes at a price. The notion that it can do some magical things that a "simpler" system like Scratch cannot is somewhat inaccurate if the criteria is the final product and not necessarily how quickly one created it. And although Autodesk will ultimately support R3d files directly, that support is not really there in a particularly complete form today (it actually is in the higher end versions of Baselight and Resolve - but once again, at a much higher price point).

As far as me having it easy, the first post house I'd worked, in Dallas, we'd used two 2" Ampex VTR-1200 (not sure about the name) for editing (yes, we did use the razor and the developing solution for edits occasionally as well) with another RCA TR-600 thrown in for good measure for back up. For color camera we had used couple of Norelco PC70? I can't remember that name either, I guess I'm really dating myself

Good lord, you're as much of an old fart as I am :crying: - and I will admit to doing spot dubs on 1200C's at one time..

Do we know each other?

 

[stephan orlandic]

Now, let's talk about being able to apply 12 layers of primary grades, HSL and RGB curves, keys with garbage masks and just simple geometries on top (while tracking), bunch of effects, grain reduction, sharpening etc? Can you see it all playing back in real time? The answer is yes for Lustre and no for Scratch. And I'm just getting started Anyway, you get my point.

This thread went little bit offtopic,
but just to add some facts.

As Lustre owner I must ask You :
WHAT RESOLUTION HAVE YOU BEEN WORKING WITH ?
1k ? SD ?
Just put a grain/blur/gold/.../, or a key/geomettry with blur on anything
like 2k or above and there is absolutely no REAL TIME.
Even on latest machine (xw8600/fx5600).
12 secondaries RT of high res grading?
Not even in a dream.

"Sharpening" ?
You must have been thinking of something else ?
No sharpening yet.

Head to head on almost the same machine,
Scratch performs bit faster when using 10bit DPX files,
using similar set of primary and secondary tools.
Most Lustre facilities grade in 1k proxy to achive heavy realtime grading.

Regards

 

[jake blackstone]

This thread went little bit offtopic,
but just to add some facts.

As Lustre owner I must ask You :
WHAT RESOLUTION HAVE YOU BEEN WORKING WITH ?
1k ? SD ?
Just put a grain/blur/gold/.../, or a key/geomettry with blur on anything
like 2k or above and there is absolutely no REAL TIME.
Even on latest machine (xw8600/fx5600).
12 secondaries RT of high res grading?
Not even in a dream.

"Sharpening" ?
You must have been thinking of something else ?
No sharpening yet.

Head to head on almost the same machine,
Scratch performs bit faster when using 10bit DPX files,
using similar set of primary and secondary tools.
Most Lustre facilities grade in 1k proxy to achive heavy realtime grading.

Regards

Yes, I agree that we got quite a bit off topic. May be we should start one for colorists?
I grade mostly in HD with some 2k in between all done with proxies. This way Lustre provides a good interactive experience. Real time playback? I can see how you may misunderstood my point. Let me explain.
Whenever client asks me to see the playback of the grade in real time or when I ned to check, if my keys are clean or for any other reason, there are three options with Lustre.
1. Cache playback. Obviously playback size is limited by 4Gb of RAM in Win 32. Just one button push.
2. Slave render. Perform the grade, move on to the next event, come back and look at P(rint)
3. Background render
But I'm sure you'd already knew this. My point is, options are there and very easy to use and set up, unlike Scratch, which doesn't even offer any. Baselight has an even slicker way of real time playback. Just play it back it once and it is rendered in cache, but it has no background render, like Lustre.
Sharpening? In the pinch just do this:
Adjust the amount of mix between the original image and the unblurred
image. By default, this value is set to 1. If you set the value to zero, no mix occurs and you see only the original image. Setting the mix to a negative value sharpens the image.

This is from Lustre user's guide:beer:

 

[jake blackstone]

I agree with you about the questionable documentation. But documentation in itself does not impact the abilities of the product. If that were the case, Sony would have been out of the professional end of the business years ago



While this is all true, these all fall under the category of "convenience features." Do they allow you to do more in less time? Yes. Personally, I don't use keyers very much in grading because they inevitably create either noise or edge problems - I tend to use either tracked shapes, curves, or vectors. However, I know that a lot of colorists do use them, so, yes, a better and simpler keyer is an advantage.



Well, maybe for you. It's never been particularly important for me. However.... You actually can do this in Scratch, by creating a custom Tray with the shots you want to group. Any trim you do on any one of them can be applied instantly to all of them. Same function, different implementation. And, yes, not as convenient or obvious as on Lustre, but available nonetheless.



True. Also true - this is another convenience feature, not something that provides more precise color correction.



Maybe on YOUR Lustre. I can show you a number of Lustre systems that it would not be true, at least not without a number of background rendering nodes.

I am certainly not arguing that Lustre (or Baselight, or Resolve) is not an excellent product. It is. But everything comes at a price. The notion that it can do some magical things that a "simpler" system like Scratch cannot is somewhat inaccurate if the criteria is the final product and not necessarily how quickly one created it. And although Autodesk will ultimately support R3d files directly, that support is not really there in a particularly complete form today (it actually is in the higher end versions of Baselight and Resolve - but once again, at a much higher price point).



Good lord, you're as much of an old fart as I am :crying: - and I will admit to doing spot dubs on 1200C's at one time..

Do we know each other?

Color grading traditionally always been the most expensive part of the post process. As a result, clients are very sensitive to the speed of execution. If I can finish the job in a third of the time with the same level of quality, where do you think clients will want to do it? Heck, you can successfully grade in FCP, but would you?

 

[M Most]

Color grading traditionally always been the most expensive part of the post process. As a result, clients are very sensitive to the speed of execution. If I can finish the job in a third of the time with the same level of quality, where do you think clients will want to do it? Heck, you can successfully grade in FCP, but would you?

I've never found the difference to be nearly that dramatic (1/3 the time??). Perhaps a 10 hour job can become an 8 hour job, but even that's pushing it. The bigger question is what's accomplished in a rather equal amount of time. If one can get through a second pass, with creative area isolations and lighting enhancements when desired, and do a bit more of a "polish" in a relatively equal amount of time, that's where I think the difference between systems begins to pay off. Besides, if the room is being billed at a higher rate - typical for a more expensive system - one enters the session with different expectations. Either they expect the job to take a bit less time, justifying the higher rate, or they expect to get more done in the time allotted. Or, more likely, they expect a more talented colorist to be in the chair and offer more creative options and a degree of sensitivity to the material and the story that justifies the higher rate. Either way, I really don't see this as a pure speed contest, particularly because in my experience, speed on any coloring job is determined primarily by two factors: how many cuts are in the show, and how consistent the original photography is. Oh, and one more: the client factor.

 

[M Most]

My point is, options are there and very easy to use and set up, unlike Scratch, which doesn't even offer any.

Yes, it does. Push one button and it will render the current shot. Or select a range and it will render the range. It will then use those rendered frames for in place playback unless you change something.

Baselight has an even slicker way of real time playback. Just play it back it once and it is rendered in cache, but it has no background render, like Lustre.

Scratch can do exactly the same thing. It is a menu selection.

 

[jake blackstone]

Yes, it does. Push one button and it will render the current shot. Or select a range and it will render the range. It will then use those rendered frames for in place playback unless you change something.



Scratch can do exactly the same thing. It is a menu selection.

I didn't know that, but how would I know it? It's not in the user manual, that I had read Regardless, from very beginning of this conversation I had stated, that Scratch is an excellent conforming and a decent grading system. If it works for you, great. I personally find it particularly lucking in the depth of tools feature set, but that's me...
So, how do you set Scratch for the render/playback swap? Which menu do I use to set it up?

 

[Pavankumar]

Hi, We can do the confirm, Color Correct, Film Out up to first print using Scratch . We just finished a movie, 'Where are you Sophia?', that is selected for Cannes film festival. Contact us with more details, and let us explore what can be done.

 

[Pavankumar]

Hi, We can do the confirm, Color Correct, Film Out up to first print using Scratch . We just finished a movie, 'Where are you Sophia?', that is selected for Cannes film festival. Contact us with more details, and let us explore what can be done.
kadapa.pavankumar@gmail.com
+91-9444524922

 

[Ido Karilla]

Comparative print test rec. 709 vs. log DPX's

Comparative print test rec. 709 vs. log DPX's

I have recently done a side by side print test of the same shots graded :
1. Linear on a calibrated rec. 709 monitor, those shots where transformed to log by the custom LUT of the Arrilaser by an independent lab that suggested this workflow. A workflow they practice on daily basis.
2. As log DPX's on my grading monitor with a LUT I got from Evangelos Achillopoulos and printed by his facility in Greece.

The tests were projected to 3 productions that where considering the Redone for a feature film and where debating witch process to follow for their film prints.

Although I was convinced that the LOG way will result in better image I was astonished by the incomparable prints. Black levels, clipped details in the whites, color spread, etc..
At the time of the projection there where two leading DP's and 3 experienced producers that unanimously decided to take the log way although the extra costs involved.

 

[jake blackstone]

I have recently done a side by side print test of the same shots graded :
1. Linear on a calibrated rec. 709 monitor, those shots where transformed to log by the custom LUT of the Arrilaser by an independent lab that suggested this workflow. A workflow they practice on daily basis.
2. As log DPX's on my grading monitor with a LUT I got from Evangelos Achillopoulos and printed by his facility in Greece.

The tests were projected to 3 productions that where considering the Redone for a feature film and where debating witch process to follow for their film prints.

Although I was convinced that the LOG way will result in better image I was astonished by the incomparable prints. Black levels, clipped details in the whites, color spread, etc..
At the time of the projection there where two leading DP's and 3 experienced producers that unanimously decided to take the log way although the extra costs involved.

Rec 709 is a color space, LOG is a gamma description. They are not the same.
You needed to compare LOG vs LIN in the same color space, such as DCI or film, than it wold have been a valid test. But at the same bit depth LOG will always work better, than LIN... BTW, why does LOG have more costs involved? Come to think of it, you could do your own LIN to LOG conversion for the grading and then deliver LOG files for the filmout, so you wouldn't have to pay to the lab to do the LIN to LOG conversion. I think it it's actually cheaper this way...

 

[M Most]

Come to think of it, you could do your own LIN to LOG conversion for the grading and then deliver LOG files for the filmout, so you wouldn't have to pay to the lab to do the LIN to LOG conversion. I think it it's actually cheaper this way...

I doubt that any lab charges for such a thing as it's really just a matter of loading a different LUT in the Arrilaser. There is no separate conversion pass.

I think it all depends on what your primary deliverable is. If it's film, you're better off working in a film color space - achieved by a combination of log source and a print preview LUT on the viewing system - not only because it gives you overall ranges that are more appropriate, but because it gives you color results that are accurate, predictable, and proper. If, however, your primary deliverable is not film, it it not necessarily the best approach, because it does, by its nature, restrict the color pallette a bit (particularly in terms of saturation), and it also applies a toe and knee characteristic that may lead to a flatter image on video deliverables than one might like. That's one of the reasons why video deliverables from high end DI houses often go through a "tweak" pass after being processed with a film targeted HD LUT baked in. For those who want the images to "look like film," though, the log/print preview LUT approach does accomplish that. And there's always the option of using Redlog, but grading without a conversion LUT - which gives you, to some degree, the best of both worlds for a video deliverable, namely an undisturbed color pallette, but with a knee for highlight compression that can make for some nice, unclipped images.

Many choices.