Welcome to our community

Be a part of something great, join today!

  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Real camera tests...

Status
Not open for further replies.
I'd be curious to see some sort of workflow survey in the feature and television industry... in particular, since anyone shooting ProRes on an Alexa, or HDCAM-SR tape on an F35, or R3D on Reds, etc. -- and in Log or RAW to boot -- and since 90% of the editing rooms use AVID (that's a wild guess), then everyone needs to deliver Rec.709 gamma footage in the DNxHD codec for offline editing... so the question is:

Is there any time difference between converting ProRes 4444 LogC vs. 4K R3D RAW to DNxHD with Rec.709 gamma? Are there any cost differences? Can all the big post houses in the major production cities do either type of conversion? If the answer is that the time and costs are the same, and all the post houses can handle either, then it's a myth that shooting ProRes 4444 HD on an Alexa is somehow more post-friendly than shooting 4K R3D RAW on a Red camera.

I'm going to start off by saying that manufacturing offline dailies is not the only step in a post workflow, and not the only one affected by the choice of capture format. I'll pick up on that in a moment.

Transcoding either format can be done by essentially every post facility in Los Angeles, but the question of time and money is not that straightforward because different facilities approach it differently. Some use the "DIY" approach of transcoding using Red software on individual machines. Others use purpose built dailies systems like Colorfront, MTI, Assimilate Scratch, and others. Still others use in house software managing other off the shelf solutions. It's very rare at a typical facility of size that only one project is going through at any one time, and it's tough to really say how long transcodes take when they're running on multiple render boxes under a render manager, processing files in a round robin fashion, and especially when the process also requires shot logging and double system sound synching, as it usually does. It is true that Red files really require a $5000 card to process in real time, something ProRes files do not. So that is a cost issue for some. And Red files often require one to pay attention to the camera settings in order to yield proper results, something also not necessary with Alexa Prores files. So those are issues. Sometimes.

Now, on the back end, it's a bit trickier. With the advent of Avid MC6 (and for some companies, even prior to that.....), ProRes files can be conformed with a simple relink process, allowing for a lot more efficiency and more accuracy on shows cut on Avid - which in the network television world is almost all of them. So conforming goes from being an 8 hour process with a lot of reverse engineering and recreation of effects to a 2-3 hour process with none of that, mostly just checking to make sure the right files are relinked. In television, that is extremely significant, and not easily achieved with Red files. Now, having said that, I worked at a facility in which we finished Red shows using native Red files, conformed on a Baselight and sent directly to grading on the same platform. And that worked pretty well, but we still had to recreate all Avid effects, and when keyframes were involved, we had to drop down to HD transcodes and process them in the Avid, then put them into the conform. Since many, if not most, shows today have even more of those kind of things, the logic of conforming on anything other than an Avid product becomes a bit suspect. And although R3D files can be read and used for relinking on Avids, it's not nearly as straightforward or efficient as doing it with ProRes files that are already HD resolution, already in log space, and require no debayering following the conform. So for television, I would answer your questions by saying that when all steps in the process are considered, yes, the ProRes files are a bit more post friendly due to their being less costly to deal with based on time and equipment needed to do it effciently.

That doesn't mean they're better, though....
 
I highly doubt a big corporation will authorize a camera test in a month. Too much red tape. I think RED should maybe delay it a month if they're interested but working through bureaucratic red tape and if they aren't interested then like Zacuto just find DPs really comfortable with each camera.

I do think the apparent motivation is due to the fact that RED didn't participate in Zacuto's tests... which seems to be RED's fault not Zacuto's. But whatever, more tests the better even if the motivation seems suspect.

The argument that RED's workflow is slow is only true if you're shooting feature film documentaries. I can render an entire TV spot on a single machine without a red rocket in less than 10 minutes. And you can always record through SDI to a Pix or Samurai or Hyperdeck plus RAW for backup.

There certainly is the potential for some bias in the test though if RED lights the scene to fit within RED's capabilities. On the other hand as I said in the Zakuto thread it makes sense to let each camera tweak their scene to fit the camera's dynamic range as an additional test.

It would be nice if each camera company could participate since they could spin the test to their own advantages. e.g. Arri point out how quiet their fan is. Give each company 2 minutes of spin time to sell the audience on what they should know beyond the image.
 
The argument that RED's workflow is slow is only true if you're shooting feature film documentaries. I can render an entire TV spot on a single machine without a red rocket in less than 10 minutes. And you can always record through SDI to a Pix or Samurai or Hyperdeck plus RAW for backup.

You will be recording a HALF DEBAYERED signal. Not good for online edit.
 
One important thing to include in any 'real life" tests is color correction looks (at least 30 corrections) and effects. The footage associated with these cc/ effects should be held up to a microscope. It is a basic fact that good footage includes these tools in modern day use.
 
SDI status...

SDI status...

I am under the impression that the next big firmware update, which based on Jarred's posts during NAB should be in the next few days, will offer us a proper SDI output from a full decode with no black bars and presumably either a SMPTE 259 or 274 compliant signal. Anyone who knows for sure, please post. I am also operating under the assumption that to get the Epics operational they did use a 1/2 debayer for the monitor output out of the gate, with plans to upgrade that down the road - which, IIRC, should be soon.

Perhaps Jarred or Jim will clarify, or perhaps they will wait until the new firmware is ready to release, that's obviously up to them.

Cheers - #19
 
Moving target

Moving target

(snip)
Now, on the back end, it's a bit trickier. With the advent of Avid MC6 (and for some companies, even prior to that.....), ProRes files can be conformed with a simple relink process, allowing for a lot more efficiency and more accuracy on shows cut on Avid - which in the network television world is almost all of them. So conforming goes from being an 8 hour process with a lot of reverse engineering and recreation of effects to a 2-3 hour process with none of that, mostly just checking to make sure the right files are relinked. In television, that is extremely significant, and not easily achieved with Red files. Now, having said that, I worked at a facility in which we finished Red shows using native Red files, conformed on a Baselight and sent directly to grading on the same platform. And that worked pretty well, but we still had to recreate all Avid effects, and when keyframes were involved, we had to drop down to HD transcodes and process them in the Avid, then put them into the conform. (snip)

All due respect to your valuable notes and extensive experience, I do question one premise; that due to ubiquity the current file handling capabilities of the AVID platform define the time frame realities of each format. For starters, native handling of R3D files with the ability to create RGB (or YUV) viewing proxies IRT and to render out 1080 or 2K RGB files of any media selected for use in better than 2x RT is already available on competing tools and I can't imagine its very far away on AVID. In any case, due to the rapid pace of format support advancement I think you have to consider the "status quo" a fast moving target which means that conclusions drawn from such a basis may have a very short shelf life.

I am confident that you understand this issue quite well Mike. I further stipulate that for many readers your depiction of how these format choices impact today's workflow/format choices is a very valuable data point. Just wanted to add the perspective that due to the burgeoning market share of professionals shooting R3Ds, and several other unique formats, that companies like AVID are quite likely to facilitate importing, file handling and metadata operations for all popular formats in a such a way that the original format choice has no noticeable impact on functionality or time frames.

Cheers - #19
 
All due respect to your valuable notes and extensive experience, I do question one premise; that due to ubiquity the current file handling capabilities of the AVID platform define the time frame realities of each format. For starters, native handling of R3D files with the ability to create RGB (or YUV) viewing proxies IRT and to render out 1080 or 2K RGB files of any media selected for use in better than 2x RT is already available on competing tools and I can't imagine its very far away on AVID. In any case, due to the rapid pace of format support advancement I think you have to consider the "status quo" a fast moving target which means that conclusions drawn from such a basis may have a very short shelf life.

I am confident that you understand this issue quite well Mike. I further stipulate that for many readers your depiction of how these format choices impact today's workflow/format choices is a very valuable data point. Just wanted to add the perspective that due to the burgeoning market share of professionals shooting R3Ds, and several other unique formats, that companies like AVID are quite likely to facilitate importing, file handling and metadata operations for all popular formats in a such a way that the original format choice has no noticeable impact on functionality or time frames.

Cheers - #19
Hope you don't mind me jumping in on this one Blair. The first iteration of Media Composer 6 and the 64-bit infrastructure is in its early stages. The plan is to optimize it over the next few revisions, and presumably implement some of the features you are asking for. At the moment, many Avid shows are edited in SD, 720P, or 1080 in an offline manner and then conformed elsewhere in whatever deliverable is needed.As time moves on, there will be demand for more to happen inside MC. Look what they are doing with working with Baselight. But for now, I'm not sure they are at that breaking point yet. I think they are trying to stabilize the transition over to 64-bit and then will add features that people are requesting. That's my take on it anyway. Mike may have some more insight into this.
 
I'm going to start off by saying that manufacturing offline dailies is not the only step in a post workflow, and not the only one affected by the choice of capture format. I'll pick up on that in a moment.

Transcoding either format can be done by essentially every post facility in Los Angeles, but the question of time and money is not that straightforward because different facilities approach it differently. Some use the "DIY" approach of transcoding using Red software on individual machines. Others use purpose built dailies systems like Colorfront, MTI, Assimilate Scratch, and others. Still others use in house software managing other off the shelf solutions. It's very rare at a typical facility of size that only one project is going through at any one time, and it's tough to really say how long transcodes take when they're running on multiple render boxes under a render manager, processing files in a round robin fashion, and especially when the process also requires shot logging and double system sound synching, as it usually does. It is true that Red files really require a $5000 card to process in real time, something ProRes files do not. So that is a cost issue for some. And Red files often require one to pay attention to the camera settings in order to yield proper results, something also not necessary with Alexa Prores files. So those are issues. Sometimes.

Now, on the back end, it's a bit trickier. With the advent of Avid MC6 (and for some companies, even prior to that.....), ProRes files can be conformed with a simple relink process, allowing for a lot more efficiency and more accuracy on shows cut on Avid - which in the network television world is almost all of them. So conforming goes from being an 8 hour process with a lot of reverse engineering and recreation of effects to a 2-3 hour process with none of that, mostly just checking to make sure the right files are relinked. In television, that is extremely significant, and not easily achieved with Red files. Now, having said that, I worked at a facility in which we finished Red shows using native Red files, conformed on a Baselight and sent directly to grading on the same platform. And that worked pretty well, but we still had to recreate all Avid effects, and when keyframes were involved, we had to drop down to HD transcodes and process them in the Avid, then put them into the conform. Since many, if not most, shows today have even more of those kind of things, the logic of conforming on anything other than an Avid product becomes a bit suspect. And although R3D files can be read and used for relinking on Avids, it's not nearly as straightforward or efficient as doing it with ProRes files that are already HD resolution, already in log space, and require no debayering following the conform. So for television, I would answer your questions by saying that when all steps in the process are considered, yes, the ProRes files are a bit more post friendly due to their being less costly to deal with based on time and equipment needed to do it effciently.

That doesn't mean they're better, though....

What if the RED RAW were transcoded to Redlogfilm ProRes4444 and these used in exactly the same workflow as the Alexa Prores? I know there is a lot of push from RED to stay RAW as long as possible but right now for TV episodic it seems like this would work very well, especially if you are comparing it to an Alexa workflow.
 
Is there any time difference between converting ProRes 4444 LogC vs. 4K R3D RAW to DNxHD with Rec.709 gamma? Are there any cost differences? Can all the big post houses in the major production cities do either type of conversion? If the answer is that the time and costs are the same, and all the post houses can handle either, then it's a myth that shooting ProRes 4444 HD on an Alexa is somehow more post-friendly than shooting 4K R3D RAW on a Red camera.

You can record in DnxHD right now on Alexa if you want. Currently 4:2:2 but looks like support for 444 is coming...

https://alshop.arri.de/catalog/alexa
 
I'd also like to include moving shot tests and hard moving. None of this looking at charts sitting on a tripod stuff. It has to be true to real life torture tests.
 
I also think no guns or knives should be allowed at the tests. Baseball bats and knuckle dusters should be ok though.
 
What if the RED RAW were transcoded to Redlogfilm ProRes4444 and these used in exactly the same workflow as the Alexa Prores? I know there is a lot of push from RED to stay RAW as long as possible but right now for TV episodic it seems like this would work very well, especially if you are comparing it to an Alexa workflow.

There are some shows (well, at least one that I know of) that are doing DNx36 with color correction for dailies, and then doing a second render to DNx175X (10 bit) without color correction - but with Redcolor 2 and RedlogFilm settings - immediately after that. The 175X files are relinked to the original timeline for the "conform." That is a pretty good way of doing things, arguably better than going to ProRes (which is non-native to Avid even though it can read it directly using AMA). In fact, on a show that we (Level 3 Post) did this season ("The River") we transcoded everything to 175X for simplicity of conform. On that show, they used at least 7 different cameras all recording different formats and codecs, so it was a particularly good way of simplifying the workflow in post. It is an extra step, though.
 
There are some shows (well, at least one that I know of) that are doing DNx36 with color correction for dailies, and then doing a second render to DNx175X (10 bit) without color correction - but with Redcolor 2 and RedlogFilm settings - immediately after that. The 175X files are relinked to the original timeline for the "conform." That is a pretty good way of doing things, arguably better than going to ProRes (which is non-native to Avid even though it can read it directly using AMA). In fact, on a show that we (Level 3 Post) did this season ("The River") we transcoded everything to 175X for simplicity of conform. On that show, they used at least 7 different cameras all recording different formats and codecs, so it was a particularly good way of simplifying the workflow in post. It is an extra step, though.
This is a workflow I use and it does work well.
 
Cute, but say Sony takes this seriously, and brings a camera to the test with their future sensor in it ? They know a thing or two about sensors.
You never know.

When there's a Dragon involved people tend to run.

- Paul
 
Status
Not open for further replies.
Back
Top