RED Workflow Rant

[Nook Kim]

I'd would like to chime in here for what it's worth. I have an established workflow for my color suite which I have been successfully using since the first day of 2009. I admire what Red has achieved in the industry, and I enjoy working with their camera even with bugs and whatever.

My thing here is something like this.. I "purchased" a Red Rocket as soon as it was available. For the first few months or so, I could understand its beta status since it sped up my transcoding at least. I was really hoping from the day one I got the card, it would be stabilized by like six months after. Now I look at my RR card and Redcine-X and have my opinion. Everytime I upgrade RC-X I really hope not to see the statement like "This driver is in the beta testing...we do not recommend this product to be used in professional environment..." I wish I could be using my RR with RC-X in my professional environment reliably someday. Last week, I found out that my QT umcomp output and QT 444 output had gamma difference, for instance. Of course, finding things like this gives me some advantages over other facilities, but.. I'd love the product that I paid to use come out of its beta mode.

 

[Jeff Kilgroe]

The RedcineX software is incredibly buggy. NONE of the DiTs i have worked with have been able to achieve realtime transcoding to Prores of RED 4khd footage as RED advertises with the red rocket card.

RED does not advertise this. They advertise real-time debayer at full quality. And the card most certainly does this. The problem with Quicktime encodes is that they DO NOT HAPPEN ON THE ROCKET HARDWARE. When transcoding, the card does a full debayer and leaves the uncompressed frames into a buffer for the CPU-bound software to encode to the desired format. If you want faster Quicktime / ProRes encodes, and we all do, then talk to Apple and/or build a faster computer. The ProRes encoder is still bound to a single CPU and has a lot of room for improvement.

The application frequently crashes. Sometimes it renders out a file without the look applied even though it displays the file and plays it in the app with the look. Honestly it's a huge pain in the ass when trying to work efficiently and RELIABLY.

Have not seen that issue with looks not getting applied and the latest builds are pretty stable. What GPU is in the system where the look issue occurs? Have you notified RED? Sent them log files? I ask about GPU because RCX does make use of the GPU for some functions, color and look functions mostly.

The color system is overcomplicated for most applications. I love the control but why call a fine ISO adjustment FLUT?

Because FLUT is more than just a fine ISO adjustment. It also has a weighted luma and gamma component. You can crank up the FLUT control and it does not clip your highlights, for example, it just compresses them.

Why not simplify all terminology and provide easy workable reliable solutions? I think a lot of people are attracted to the 5D because of it's simplicity.

Hmmm... I suppose it's subjective, but I don't see what is so complicated here. I think most of the issue comes back to the continuous changing and the minor learning curve. People do not like to learn new things I guess. As for the 5D, I don't understand the continuous comparisons or all the claims about it being so easy. The 5D/7D are a pain in the ass when it comes to working with audio. They acquire in a consumer-grade distribution format, line skipping, aliasing, etc.. These HDSLR cameras cause more headaches in post than RED ever has for us. Any serious compositing and FX work is a nightmare with footage from a 5D/7D. I've seen a lot Canon DSLR footage over the past year, and have shot a fair bit myself, and frankly, it sucks. Shooting with these HDSLRs is only easier from the point of view that the footage won't change much, if any, from acquisition.

Easy to understand and familiar terminology

Hmmm... What are you asking for here? All of RED's terminology within the post realm is pretty straight-forward. Sure, there's REDcolor, REDgamma and the older REDspace, but they have to call it something. If not that, then what? There's R3D files that contain RED RAW data. Then there's FLUT... So what is it you're looking to clarify?

a very easy workflow with final cut

Talk to Apple. It's their software. The Canon FCP plugin is not even in the same league as to what is required for RAW R3D clips. In fact, the Canon FCP plugin is not even necessary.

WYSIWYG results from shooting RAW? On an uncalibrated camera monitor? The closest you're going to get to that is to have a calibrated directors' monitor in a controlled viewing environment that receives an SDI or HDMI feed from the camera with a look applied. And then, camera interpretation is one thing. How other post tools interpret RAW and apply color spaces is up to those application developers.

I enjoy shooting RED and raw but the software difficulties and color inconsistencies have grown tiresome.

Do you shoot and process RAW stills? How do you deal with inconsistencies with those RAW formats? You do realize that most every application out there differs from another in how they interpret RAW and apply color space, right? Or do you stay confined to one application or color system, such as Photoshop with Adobe Camera RAW, so there is no disagreement to worry about?

Which leads to another observation here. I'm seeing that most of the people with the biggest reservations or issues with RED post workflow are those who feel they need to (or just want to) remain confined within a particular family of applications. Seeing a lot of dissent over RED workflow from the Final Cut crowd. Understandably so... FCS sucks for RED workflow right now without doing a massive up-front ProRes transcode or relying on third-party tools and/or scripts.

Lots of Final Cut people wanting a better workflow experience and blaming RED for this. They need to talk to Apple.

On camera creation of ProRes or high quality (useable for broadcast finish) files is a must have for REDs next cameras to compete and be attractive in the marketplace.

There is the H.264 module coming. I suppose a ProRes module is possible, but if it gets done remains to be seen. Same with other formats. While I see the usefulness and application for such modules, I also fear that many will look at them as a shortcut or a "good enough" solution for shooting with RED. People will opt to go straight to H.264 right out of the gate and not touch the RAW.

ProRes? Give me raw any day. I'm sure I will get flamed for saying this, but in my opinion, if you are serious about your images, you shouldn't be shooting video on a DSLR, and you shouldn't be grading in ProRes. Super8 cartridges were a lot more convenient that Super35 magazines too... convenience shouldn't trump quality (and frankly, with native R3D support in Adobe and Avid, you can have the best of both worlds).

I fully concur.

Last week, I found out that my QT umcomp output and QT 444 output had gamma difference, for instance. Of course, finding things like this gives me some advantages over other facilities, but.. I'd love the product that I paid to use come out of its beta mode.

The QT gamma issues are out of RED's control. This is entirely an Apple problem. The Rocket card transfers the full quality, uncompressed frames over to the Apple QT Codec, which in turn creates the QT files. It is responsible for compressing/butchering the gamma. I suppose RED could attempt to compensate by offering a gamma compensation for QT output. However, I could see that as ultimately causing more problems.

 

[Steve Freebairn]

Does anyone think, that any of the tools mentioned (Clipfinder, PP, RC-X etc) were used on those big productions? What's the point of knowing the workflow and tools used, if those tools are WAY out of reach?

Trust me, Clipfinder is being used on some of the biggest shows. If a large show is not using it, then they've spent a lot of time making something "custom" that doesn't do everything that Clipfinder does. Hans has some new tricks up his sleeves, if you're not impressed now, you soon will be.

 

[Jeff Kilgroe]

Trust me, Clipfinder is being used on some of the biggest shows. If a large show is not using it, then they've spent a lot of time making something "custom" that doesn't do everything that Clipfinder does. Hans has some new tricks up his sleeves, if you're not impressed now, you soon will be.

I can vouch for this too. Clipfinder is used heavily on productions of all levels. Same with REDCINE-X.

There are always kinks in most every workflow and no one software solution does it all, nor does what it does perfectly. A workflow consists of using whatever tools you may have in your toolbox to get from your RED RAW acquisition to a finished product. Finding a good workflow for the project at hand often means changing tools or combining tools in new ways to accomplish what you need to.

 

[Tim Whitcomb]

A large part of what I do for a living is design, modify, and implement workflows for projects - both studios and independents. And while many of the good points brought up in this thread contain validity, I consistently observe a lack of understanding of RED that is letting people down - not the RED itself.

Workflows are fickle. I don't even like that term, "workflow." They are like cars of a train - some have a dozen, some have a hundred. They move and change, and require constant modification. In fact, in working on hundreds of RED projects, it's extremely rare for my team to implement the exact same workflow twice. I kind of think of it like cracking an electronic safe:
When you find a number that works, you leave it. But concurrently you're trying out new numbers...one after the other...until you find another that works. When you hit a number that doesn't work, you make a note, and make a change.
As each component of a workflow falls into place, make a note as to the impact and evaluate its pros and cons. Use this combination as long as you can...that is until you come across a new safe requiring a new combination...

The second largest part of what I do for a living is educate (which I enjoy). For many people reading this thread, I ask that you take a serious look at attending REDucation this coming August. REDucation is not a week in which you will spend money reviewing what you already know. I promise (former REDucation attendees, please chime in REDucation touches each and every one of the issues brought up in this thread, discusses them, demonstrates them, and solves them.

Answer me this: If I were to publish some of the workflows I've implemented on shows, (such as outlines, budgets, or flowcharts) would you find any benefit to reading them? For example, the comment about RED having 3 color sciences - in professional film and video, there are way more than 3 ways to interpret colors. I would die if I only had to worry about 3 ways to debayer So if I were to publish how we manage RED projects amidst the issues that have been brought up here, would that help? What if I posted some of the key workflow points we're doing on Pirates 4, or Social Network, would anyone find that useful for the community?
(maybe there is some offline REDucation we can do here on RU???)
m

Hi Michael! I third the requests and welcome your input. As a REDUCATION graduate last year... you guys taught us THE CABOOSE EFFECT and its been an invaluable way to approach working with RED RAW in post.

That is simply, start with your deliverable, and then engineer the workflow BACKWARDS from there. Sounds simple. and it is. Are you film out? Are you finishing on Blu Ray? DVD? Broadcasting on an antiquated SD cable channel?
Start with your deliverable, and work backwards is the best advice ever!

Yes, we are spoiled by the PABLO, but a lot of our indie clients cant afford that suite and work mostly in FCP and COLOR, the most problematic software to work with RED imho. What we ALL can do politically, is make sure the BLAME is falling in the right place. RED is not responsible for MOST of the FCP issues. Its all on APPLE. KNOWLEDGE IS POWER.

I strongly endorse REDUCATION and can tell you I have recouped that investment 4X easy in the past year! Sell your EVF (they are selling for full price right now) and take the AUGUST class!

 

[Andrew clemson]

Id take REDucation again if I had the chance. It was a great laugh.

 

[Curran Giddens]

Last week, I found out that my QT umcomp output and QT 444 output had gamma difference, for instance. Of course, finding things like this gives me some advantages over other facilities, but.. I'd love the product that I paid to use come out of its beta mode.

By QT 444 do you mean ProRes? This is an Apple issue, not a Red Rocket issue.

 

[Petri Teittinen]

I have to throw in my $0.02 as I've been struggling with workflow issues for the past few weeks now. I'm running a Win7 64-bit, and being more of a hobbyist than a proper filmmaker I can't afford really high-end solutions. My choices at the moment include Sony Vegas 9 Pro, Adobe Premiere Pro CS5 and RC-X. But here's the snag:

1) Vegas 9 Pro supports FLUT and new colour science, but has no support for RRocket or CUDA acceleration.

2) Premiere Pro CS5's new R3D importer doesn't work properly in my system. I can't import footage shot using latest firmware to PPro 5.0.1 unless I go back to earlier R3D importer - which does not support FLUT or new colour science.

3) Even the latest release (b252) of RC-X is very shaky. While Vegas 9 and Premiere Pro have no problem accessing clips from an external SAS array, RC-X manages to crash the SAS adapter's driver, causing a brutal BSOD.

I have a CUDA-capable display adapter and a RRocket in my system but since the only working FLUT-supporting solution is Sony Vegas 9 Pro, neither is of any use to me at the moment. This is quite frustrating.

 

[Gunleik Groven]

Clipfinder is an integral part of a lot of heavy iron workflows. Because it is good and it works. Rocketcine X 1740 has also been dead stable when treated right... -

As to changing from FCS, that is an easy task in a small organisation, but not in an organisation where every software needs to be tested for compatability and workflow issues with the rest of the house in a larger environment where RED shoots are only a part of the production. Everything needs to plug into the organisation and typically even upgrades aren't done tested untill 3 months after their release. Jumping to AVID PP or anything else because it works best right this moment is a non-issue...

RCX is for the time being a non-issue.
But I am sure that will change.

Testing a couple of Resolves, on the other hand... -

 

[jake blackstone]

I can vouch for this too. Clipfinder is used heavily on productions of all levels. Same with REDCINE-X.

There are always kinks in most every workflow and no one software solution does it all, nor does what it does perfectly. A workflow consists of using whatever tools you may have in your toolbox to get from your RED RAW acquisition to a finished product. Finding a good workflow for the project at hand often means changing tools or combining tools in new ways to accomplish what you need to.

Multimillion dollar productions can't rely on mom and pop software, no matter how useful. All top post houses, without an exception, use either Baselight, Pablo, FilmMaster, Resolve or Lustre. For those products technical support is only a phone call away, 24/7. Troubleshooting can be done remotely by the post house staff engineer or the manufacturer. They can ssh into the system and take care of the problem immediately. As part of annual support, Filmlight even does routine maintenance on the system, if needed.
I doubt Hans has the resources to provide that kind of level of support for such an inexpensive software.

 

[Eric Haase]

The color inconsistencies im talking about are monitoring the hd sdi out on the camera with a calibrated HD sdi monitor. Looks one way on the monitor. Load it into rcx through a rocket card on the same monitor with same metadata and it looks different. This means you HAVE to have SOMEONE dial in the look on a computer if you want the look you set on the camera. Its creating the look twice. For some applications, like easy dailies, it would be really nice to just use the camera metadata and have the result look exactly like it did when monitoring off the camera. I don't think his is a huge demand.

I don't want to have my DIT spending his time sending log files and helping debug super buggy software. I want him dialing in looks, managing data, and giving me the thumbs up that everything is a ok. On the last tv pilot i did we have several clips that could not be transcoded at all with rcx. They would hang the app.

Sure raw is great and it's ideal and i would rather shoot raw and grade raw etc etc. But the reality is, there are a lot of instances when it doesn't make sense and when prores is good enough. Look, even high end car commercials are shooting running footage and interior footage with the 5D. This should tell you something about the importance of image quality (acceptability) versus ease of use and cost.

The bottom line is that working with red raw and the red software has for the most part been incredibly frustrating for a lot of people. I've spent hundreds of hours becoming incredibly proficient with a lot of the solutions out there. And i dont even post the footage myself. But i have to arm myself with this knowledge BECAUSE things dont work properly so i have to explain to producers why we dont have that take we shot 2 days ago transcoded yet. I love shooting red but i can tell you that all of these inconsistencies and problems are making me love it less. I doubt i am alone in this. I just want RED to really try to simplify things and make simpler things that work without bugs and log files and crashes. I hope that the competition from the DSLR market is a challenge to move things in this direction and think about simplicity and ease of use along with quality.

 

[Gunleik Groven]

Sounds very unfamiliar, but then I avoid RCX at all costs untill it's out of beta. RocketcineX 1740 and clipfinder is an extremely unbuggy combo...

 

[Jeff Kilgroe]

Multimillion dollar productions can't rely on mom and pop software, no matter how useful.

And they don't need to, but a lot of multimillion dollar productions are using Clipfinder. A lot of them are not. It's an available tool, no reason to discount it.

I said in one of my posts above that I don't really use clipfinder. Which in turn comes from this same logic. I have built most of my workflows to be platform agnostic where possible and to not rely on mom n' pop software where possible, except tools I create myself. I have my own data management tools that I use for offloading, verification, etc.. Most of which I have built long ago for other purposes and I used them heavily during my days shooting and offloading P2 media.

I do use Clipfinder once in a while, but typically don't find I need it.

All top post houses, without an exception, use either Baselight, Pablo, FilmMaster, Resolve or Lustre.

Interesting... I was DIT on a $25M production two months back and they were, and still are, not using any of those.

Sorry, I'm just messin' with you a bit. The production is finishing on Scratch, which you forgot to mention along with some others. But your point is taken. And where many of the holes in RED post occur is in on the lower-cost end of the application chain. While there are workable solutions at all levels of production, there is definitely a lack of training and support in some areas. Your example only mentions applications or systems that are primarily for finishing. There are plenty of multimillion dollar productions that are shooting RED and are being editing in Final Cut. Workflow is not just about finishing, it is about the entire process from start to finish. Not everyone has the luxury of finishing on a Pablo, but that does not mean there aren't solutions at a lower price point.

I'm still confused how your example of the high-end finishing options has anything to do with using Clipfinder.

As for Clipfinder and Hans and support, sure... But Hans is not even attempting or pretending to offer such a solution. He's just now starting to make a few software tools that cost money. Clipfinder is a very simple tool and not to be confused with any of the systems you mentioned. I know some post houses using Clipfinder and they still finish on Scratch, SpeedGrade or Pablo. I know others that have never touched Clipfinder and don't feel they need to.

 

[jake blackstone]

Because FLUT is more than just a fine ISO adjustment. It also has a weighted luma and gamma component. You can crank up the FLUT control and it does not clip your highlights, for example, it just compresses them.

Graemme recently wrote, that FLUT and ISO are exactly the same. He explained, that traditionally ASA is adjusted in steps of 1/3 of a stop, where FLUT is in .1 of a stop. There are no other differences.
Of coarse, now I can't find the thread, where he said that...

 

[Luca Immesi]

.
I doubt Hans has the resources to provide that kind of level of support for such an inexpensive software.

All your considerations are true but I tell you this: during reducation #1 Steve was demoing Clipfinder, there was something weird with a function, I can't remember what but I wrote an email to Hans about it and he sent us a new version in less than an hour. For sure the big productions use heavy iron solutions but often Clipfinder is the missing link between editing and color correction.

 

[jake blackstone]

And they don't need to, but a lot of multimillion dollar productions are using Clipfinder. A lot of them are not. It's an available tool, no reason to discount it.

I said in one of my posts above that I don't really use clipfinder. Which in turn comes from this same logic. I have built most of my workflows to be platform agnostic where possible and to not rely on mom n' pop software where possible, except tools I create myself. I have my own data management tools that I use for offloading, verification, etc.. Most of which I have built long ago for other purposes and I used them heavily during my days shooting and offloading P2 media.

I do use Clipfinder once in a while, but typically don't find I need it.



Interesting... I was DIT on a $25M production two months back and they were, and still are, not using any of those.

Sorry, I'm just messin' with you a bit. The production is finishing on Scratch, which you forgot to mention along with some others. But your point is taken. And where many of the holes in RED post occur is in on the lower-cost end of the application chain. While there are workable solutions at all levels of production, there is definitely a lack of training and support in some areas. Your example only mentions applications or systems that are primarily for finishing. There are plenty of multimillion dollar productions that are shooting RED and are being editing in Final Cut. Workflow is not just about finishing, it is about the entire process from start to finish. Not everyone has the luxury of finishing on a Pablo, but that does not mean there aren't solutions at a lower price point.

I'm still confused how your example of the high-end finishing options has anything to do with using Clipfinder.

As for Clipfinder and Hans and support, sure... But Hans is not even attempting or pretending to offer such a solution. He's just now starting to make a few software tools that cost money. Clipfinder is a very simple tool and not to be confused with any of the systems you mentioned. I know some post houses using Clipfinder and they still finish on Scratch, SpeedGrade or Pablo. I know others that have never touched Clipfinder and don't feel they need to.

I don't have any intention to denigrate Clipfinder. I personally don't use it, but all people that use it and swear by it can't all be wrong?
I made an earlier point, that high end post houses, like Company 3, Technicolor, Sony, WB, FOX (Delux EFilm) etc. they all use the tools I had mentioned. I neglected to mention Scratch intentionally. To my knowledge and I could be wrong, none of the companies, that I had mentioned, use Scratch for actual color grading, just conforming. Scratch is a fantastic conforming tool... But I digress. Finishing is a process. If any part of that process breaks down, the whole process comes to a screeching halt. High end post facilities have redundancies built in, where they can just move into another room and pick up, where they left off. Director or DP time is too valuable to sit around, while engineers trying to figure out how to fix the software problem. The process has to be absolutely bullet proof. So, software with no 24/7 remote support, no vigorous production testing, no dice... That is all I was saying.

 

[Justin Barham]

Answer me this: If I were to publish some of the workflows I've implemented on shows, (such as outlines, budgets, or flowcharts) would you find any benefit to reading them? For example, the comment about RED having 3 color sciences - in professional film and video, there are way more than 3 ways to interpret colors. I would die if I only had to worry about 3 ways to debayer So if I were to publish how we manage RED projects amidst the issues that have been brought up here, would that help? What if I posted some of the key workflow points we're doing on Pirates 4, or Social Network, would anyone find that useful for the community?
(maybe there is some offline REDucation we can do here on RU???)
m

That would be HUGE.

 

[Jeff Kilgroe]

Graemme recently wrote, that FLUT and ISO are exactly the same. He explained, that traditionally ASA is adjusted in steps of 1/3 of a stop, where FLUT is in .1 of a stop. There are no other differences.
Of coarse, now I can't find the thread, where he said that...

That's essentially what it is on the surface, but it's also more...

There's very little point having ISO that is pure gain and hard clips the top end. This is not new to us, but making it work otherwise in a seamless and easy manner proved to be trickier than we thought, especially as we like to have the same image processing in hardware as we do in software.

So with FLUT, we give you that feature. It's gain without hard clip, which means you don't loose stops of DR for each stop of gain you go up. Instead you crush the higlights nicely and smoothly. No add 6 stops of FLUT and you're going to really crush those highlights - but none are sent to hard clip even though they may look like it.

If you do want traditional straight-line hard clip gain, use the "exposure" control.

Graeme

 

[M Most]

That's essentially what it is on the surface, but it's also more...

No, it's not. The ISO setting and the FLUT control are exactly the same except in terms of scaling. The Exposure control is a different control that does not use the FLUT math and is, I believe, downstream of the FLUT based controls.

 

[David Battistella]

With FLUT science in place it prevents clipping at any ISO, the FLUT is always engaged. The FLUT control slider is a fine tuning of ISO, like being able to get in between ISO stops.

David