How Many Of You Planning A Movie Theater Feature On The 8x Fixed Scarlet?

[John Warner]

Heya long time lurker and first post

I was just wondering after reading here for a while how many of you are planning on shooting a Theatrical Movie Theater Release on the 8x scarlet fixed, and if you guys had any opinions on the subject.

As in down rezing the 3k to a 2k digital intermediate for the final post,edit,grade and transfer to 35mm or things of that nature(sorry if this statement is inaccurate as i am a beginner

I would love to see how many of you all are awaiting this camera just for that purpose, as well as any of your ideas on workflow with a targeted theatrical release with the scarlet 8x fixed

 

[Nuka Wølk]

Hi John,

I guess most users in here would tell you: Yes, sure. But...

As we all know, the SCARLET isn't out there...yet! - and nobody(but RED) knows exactly when this camera is gonna go on the market. Also, with the HUGE demand for this camera, who really knows when this camera will actually land in your hands so you can start shooting?

So if you're a serious and responsible filmmaker(and we all are of course ) - you'd probably work out your planning from which camera you'd actually be able to get your hands on right now, and be able to do some test-shots with first (and there are a lot of good cameras out there...at least until this(SCARLET) probably will render them all out at some point). Choosing the camera you want for a certain production isn't just something you can "wing" (though I imagine someone would). You have to think down the line in terms of production workflow - and of course: budget. How will the choice affect everything else in your production. Especially if you are new to all of this(high-end) you have to realize that you commit youself to a certain workflow when you choose a certain camera - and when you go "pro", the price of the workflow reflects that. And that might not be realistic for every "indie"-production out there (unless: get real friendly with people from post houses - "wink, wink"). ;D

Nevertheless, just the idea of doing a theatrical feature on this tiny powerhouse of a camera is a very, VERY intriguing thought - I know, I'm a filmmaker myself and I can't wait to try this baby out - so I hopefully can start planning my SCARLET-projects!

 

[Randall Nott]

John,
Yes, I plan to use Scarlet to shoot a feature, but not waiting for it just for that. I think the size and portability of it will make it my go to camera for most projects - especially with the rapid growth in capacity of aquisition media it will soon be viable for event work as well.

I don't understand why you would want to down-rez. The whole point of a digital intermediate is to give you more resolution and latitude than your source footage. If raw r3d files are your source footage, you already have the most resolution and latitude you're going to get. A Red Rocket card, the free Red software, and Adobe Production bundle and I'm set. All of the major NLE's seem to have a Red workflow at this point.

Nuka,
Good points on planning, but I think if your just planning a theatrical production now, Scarlet will either be ready by the time you're ready to shoot, or renting a Red One will be an alternative you can plan on - the workflow will be relatively the same.


My biggest concerns with planning at this point are on the side of advertising and distribution, areas often neglected here that shouldn't be left to chance. Your target audience and how costs will be recouped makes all the difference in how you scale your production. A few years ago I went to the Sundance festival and followed the fates of the two dozen or so films I saw, as well several others that had gotten good buzz, over the next couple of years. The festival darlings are not necessarily what will do well. The big lesson was the higher you aim, the better your odds are. The more you can count on professionals, and less on your friends, the better your odds are. Get SAG Indie contracts for you actors - this was the biggest factor on if a film made it; who's attached to an indie project will influence who will back it. If you insist on directing, hire a cinematographer. If you insist on directing and shooting, hire an art director. If you don't need someone else's design input, you should want it anyway - the right person will inspire you. If you're not ready to make a feature, don't. There are plenty of small projects you can maximize your resources with, or participate in while learning your craft and growing your network. But if you burn out your friends and family (and finances) on a feature going nowhere, you will likely go nowhere with it.

 

[Stephen Gentle]

I expect that there will be many indie features shot on it. Hell, even some films that saw international theatrical releases were shot on DV (such as parts of 28 days later that were shot on a Canon XL1), so there may be some theatrical release features shot on Scarlet too.

I don't understand why you would want to down-rez. The whole point of a digital intermediate is to give you more resolution and latitude than your source footage. If raw r3d files are your source footage, you already have the most resolution and latitude you're going to get. A Red Rocket card, the free Red software, and Adobe Production bundle and I'm set. All of the major NLE's seem to have a Red workflow at this point.

I don't understand what you're getting at. First of all, you can't get more resolution and latitude than your source material (unless you're talking about titles and CG but not actual footage). That's locked in from when you shoot it. What a good post workflow should do is to keep all the resolution and latitude you have until you get to the end and finally produce deliverables.

As for downscaling, the point is that there is no 3K standard for distribution. At the end of your post pipeline, you're going to be making 1080p deliverables for Blu-Ray, and if you are going to cinema (be it digital or a film out) you either have to go down to 2K or up to 4K. For a filmout, it would probably always be 2K for Scarlet (4K prints are too expensive, last time I checked) and for digital, you'd have to do a comparison in a real theatre but I don't think that there'd be much to gain by going up to 4K.

 

[John Warner]

As for downscaling, the point is that there is no 3K standard for distribution. At the end of your post pipeline, you're going to be making 1080p deliverables for Blu-Ray, and if you are going to cinema (be it digital or a film out) you either have to go down to 2K or up to 4K. For a filmout, it would probably always be 2K for Scarlet (4K prints are too expensive, last time I checked) and for digital, you'd have to do a comparison in a real theatre but I don't think that there'd be much to gain by going up to 4K.

Thanks for all the info guys, and this is exactly what i was understanding, is after all the post is done the scarlet 3k footage would then become a digital intermediate 2k master and then from that 2k master become the final 35mm print.

I'm actually an audio guy by trade and have been recording, editing, and mixing various projects for many years now (nothing major though) so the thought of wearing many hats, as well as understanding the nature and complexity of multimedia based projects is not something new to me, although the thought of producing a feature is intimidating.

But i do feel with an extensive audio background and an eye(or should i say ear heh) for detail that i would like to try and take on the challenge of producing a feature, and i am very curious as to how many others are also thinking about it (no matter what their skill level may be) to make a feature film with a targeted theater release on the scarlet 8x fixed.

As i read some people say that the fixed lens wont be able to handle this or that, and i am in now way qualified to enter that debate on the technical side, but on the theory side i feel that some of the most amazing art i have ever had the privileged to see or hear has been a victim of circumstance.

 

[Benoît-Joan Clariana-Roig]

Planning some documentaries (wild life & life style). Not sure if they'll hit the big screen someday but it's good to get 3K masters. :laugh:

I'll also work on a web-series & a zero G flight film shoot is planned next fall in France. It's getting exciting more & more everyday !

 

[Alexander Ibrahim]

I may do some 3D work with Scarlet Cinema, and I'll probably use Scarlet as my go to camera for ENG/EFP style TV replacing traditional shoulder mount broadcast cameras.

That last part depends on a lot - including the adaptability of broadcast B4 lenses and the exact modules and camera grip available to conform the camera to the ENG/EFP role.

As far as your post questions, I agree with Stephen. The ideal is to stay in RAW until you create your final deliverable. There are valid reasons to step away from RAW sources (and it doesn't matter if we are talking about Red R3D or Arri .ari RAW) earlier in post ... but every such step is a compromise on image quality.

Of those compromises the one I prefer most is to do a careful grade towards final picture on the RAW, then output your editorial format. You use the flexibility on the front end of the process. This is far from ideal, and it requires you to do a lot of planning and expend unnecessary effort grading files you wont use in order to achieve some other goal. The process works, but its a compromise.

 

[Randall Nott]

I don't understand what you're getting at. First of all, you can't get more resolution and latitude than your source material (unless you're talking about titles and CG but not actual footage). That's locked in from when you shoot it. What a good post workflow should do is to keep all the resolution and latitude you have until you get to the end and finally produce deliverables.

Sorry I phrased that badly. Besides pixel resolution, there is colorspace and bit depth resolution. For example, HDV uses 8-bit yuv in a 4:2:0 colorspace and a resolution of 1440x1080. You would want to transform that to an intermediate using 10-bit yuv in 4:2:2 or 12-bit rgb in 4:4:4, while uprezing to 1920x1080, and hopefully move it out of long gop in the process. So for lower end HD cameras, you can and should uprez before doing post work. Red already has better specs, so yes, I agree, "What a good post workflow should do is to keep all the resolution and latitude you have until you get to the end and finally produce deliverables." That's the point I was trying to make.

 

[Alexander Ibrahim]

Sorry I phrased that badly. Besides pixel resolution, there is colorspace and bit depth resolution. For example, HDV uses 8-bit yuv in a 4:2:0 colorspace and a resolution of 1440x1080. You would want to transform that to an intermediate using 10-bit yuv in 4:2:2 or 12-bit rgb in 4:4:4, while uprezing to 1920x1080, and hopefully move it out of long gop in the process. So for lower end HD cameras, you can and should uprez before doing post work. Red already has better specs, so yes, I agree, "What a good post workflow should do is to keep all the resolution and latitude you have until you get to the end and finally produce deliverables." That's the point I was trying to make.

Well that's sort of right. But "sort of right" means "actually wrong."

You don't want to up-res until final delivery ... because the interpolation by necessity softens the image. Some systems compensate by artificially sharpening- which may look better but multiplies the problems. For HDV the best choice is to work with the 1440x1080 raster. Its better to have sharp accurate 1440x1080 than sloppy, soft inaccurate 1920x1080.

If you are using a camera codec that is processor intensive like HDV or h.264, you will want to convert to a better editing format, like DNxHD or ProRes 422. Doing that is a transcoding step, and actually REDUCES quality.

On most of the productions that use HDV or DSLR/AVCHD ... no one cares.

Now, if you have a rare production where ultimate image quality is desired despite the source ... then you want to use the 10 bit 4:2:2 Uncompressed codec at the camera original resolution. ( 1440x1080 in the case of HDV. )

As far as the bit depth and such ... you aren't creating anything new- but by using a higher bit depth "container" you allow a tool like FCP to work in higher precision- so all the rounding errors in the 8bit 4:2:0 math end up in. That's good.

Now, you can apply the same thinking to a quality acquisition format like R3D or ArriRaw, but you have a whole different set of problems.

For one, once you reach that level of image acquisition, you will discover that the camera designers are very smart about being highly efficient with the data storage from the sensors they are working with. They have very high bit depths already, and those bit depths are already highly optimized in comparison to what the sensor can see.

Let me say that differently. With Red and Arri cameras the engineers optimize to the limit of the technology they have available. With HDV and DSLR they engineers optimize to the limit of the target price point.

So, basically don't screw with it. Work with the R3D directly (or the .ari files) until the end. If you must change formats, OpenEXR 16 bit is the one you want to use, and you probably also want to use ACES-IIF aware color control systems/applications.

If you don't know what that means ... then definitely don't screw with it.

 

[John Warner]

So, basically don't screw with it. Work with the R3D directly (or the .ari files) until the end. If you must change formats, OpenEXR 16 bit is the one you want to use, and you probably also want to use ACES-IIF aware color control systems/applications.

If you don't know what that means ... then definitely don't screw with it.

Thanks for this, so if i may , would you suggest picking up a red rocket card and doing all of the post, editing, grading etc, in native r3d format? and then settling on output formats from the final sequence of r3ds? (i hope this makes sense)

As i was considering a transcode in ither prores or DNxHD, but i would really like to know what more experienced people would do in this situation, as there is also the proxy alternative(but i would assume this would not work for grading or would it?)

whether it hits a theater near you or just one in my local town, in this hypothetical situation i would really like to know how you guys would work with footage at 3k. As well as how many people are planning a theater feature with the 8x fixed.

 

[Alexander Ibrahim]

Thanks for this, so if i may , would you suggest picking up a red rocket card

Yes I would suggest it.

and doing all of the post, editing, grading etc, in native r3d format? and then settling on output formats from the final sequence of r3ds? (i hope this makes sense)

Yes it makes sense, and yes its exactly what the Red folks recommend.

I will say that increasingly applications are working with R3D directly and that you don't need to use the Rocket until final output.

I frankly think that the folks at Red are hoping the rest of the computer and post industries catch up so they don't have to come up with a "Rocket 2"

As i was considering a transcode in ither prores or DNxHD, but i would really like to know what more experienced people would do in this situation, as there is also the proxy alternative(but i would assume this would not work for grading or would it?)

Avoid doing a transcode at any point before delivery if at all possible.

whether it hits a theater near you or just one in my local town, in this hypothetical situation i would really like to know how you guys would work with footage at 3k. As well as how many people are planning a theater feature with the 8x fixed.

The 8X fixed is almost certain to have footage turn up in a lot of theatrical releases, and lots of broadcast. At least so long as 2K/1080p delivery is acceptable.

The good news about 3K footage is that its a lot less data than 4K or 5K, so its faster to work with if you can not afford a Rocket.

Good luck with your feature!

 

[Randall Nott]

Hi Alexander,

On most of the productions that use HDV or DSLR/AVCHD ... no one cares.

LOL, too true. I might add, "or notices."

You don't want to up-res until final delivery ... because the interpolation by necessity softens the image. Some systems compensate by artificially sharpening- which may look better but multiplies the problems. For HDV the best choice is to work with the 1440x1080 raster. Its better to have sharp accurate 1440x1080 than sloppy, soft inaccurate 1920x1080.

If you are using a camera codec that is processor intensive like HDV or h.264, you will want to convert to a better editing format, like DNxHD or ProRes 422. Doing that is a transcoding step, and actually REDUCES quality.

If all you are doing is straight cuts, this is true. But if you are working with effects, mixed footage, or hi-res plates, then the debate is whether it is better to take the upres hit at the beginning of the process while transcoding, or at the end, assuming your end result is going to be "soft 1920x1080" anyway (the curse of HDV). I've tried it both ways, more than once. My experience is taking the upres/transcode hit at the front end of the process gives me more headroom or latitude when manipulating the footage. For a better editing format, I'm using Cineform Prospect 4k (new version is Neo 4k) for 10 bit yuv 4:2:2 (and 1920x1080). Like DNxHD and ProRes 422, you do take the quality hit, but it holds up better through subsequent processing meeting their "visually lossless" claim. The final footage usually looks better than playing the tape from the camera to an HDTV, simply because cineform has a better scaling algorithm than my tv does. I also believe that Cineform has better scaling than Adobe encoder does, so the front end up-res works better for me.

As far as the bit depth and such ... you aren't creating anything new- but by using a higher bit depth "container"...

You absolutely are creating something new, you're creating the higher bit depth "container", and that is what makes all the difference in processing. And whenever you are applying an effect, you are creating new footage, literally in the case of Adobe. I would rather have that new content created at the higher resolution level; gives a better end result. Absolutely agree with all your comments regarding raw footage.

John,
The reason I brought up the process for these crappy last generation formats, is that this is what I think is influencing your impression that you need to transcode. You don't. As Alexander says, R3D to the end. The sooner Scarlet puts the nail in the coffin of these formats, the better.

...would you suggest picking up a red rocket card and doing all of the post, editing, grading etc, in native r3d format?

Yes, but the card is only necessary if you want hours of your life back. In Adobe Premier, you can take your LUTs from grading in Redcine-x and apply them directly to your clips. Adobe had a nice demo of this at NAB, and will hopefully post it on their training website.

and then settling on output formats from the final sequence of r3ds? (i hope this makes sense)

Yes. Makes the best sense.

 

[John Burnley]

I'm planning a horror feature for spring or summer of 2012. Fixed Scarlet would be my backup camera of choice for this. I say back up because at this point I don't know what Industry players may or may not come into the production. If there's Hollywood backing I'd shoot on either Epic or 35mm film (partly depending on budget) but if I need to fund this feature on independent capital I'd be very happy to shoot on Scarlet -- or Red One.

For my pre-production trailer this year, since Scarlet Fixed will not likely be available in time, then I'd go with Panasonic AF100 which should be adequate for DVD presentation. Red One would of course be another option for this.

John Burnley

 

[Kalani Prince]

I was planning a documentary for Scarlet but decided to wait till 2012ish.. In the meantime I will shoot and on-going "movie" via the YouTubes with Scarlet. This is gonna be fun ; )

 

[Paramon Ivanovich]

I would love to see how many of you all are awaiting this camera just for that purpose, as well as any of your ideas on workflow with a targeted theatrical release with the scarlet 8x fixed

yeah, feature would be nice, but i'm buying two fixed ones to film 3D weddings

 

[John Warner]

yeah, feature would be nice, but i'm buying two fixed ones to film 3D weddings

hah thats awesome!

 

[RAVI WADKAR]

I am going for
DPX RELEASE in theatres.... any suggetions

 

[Omar]

Well that's sort of right. But "sort of right" means "actually wrong."

You don't want to up-res until final delivery ... because the interpolation by necessity softens the image. Some systems compensate by artificially sharpening- which may look better but multiplies the problems. For HDV the best choice is to work with the 1440x1080 raster. Its better to have sharp accurate 1440x1080 than sloppy, soft inaccurate 1920x1080.

If you are using a camera codec that is processor intensive like HDV or h.264, you will want to convert to a better editing format, like DNxHD or ProRes 422. Doing that is a transcoding step, and actually REDUCES quality.

On most of the productions that use HDV or DSLR/AVCHD ... no one cares.

Now, if you have a rare production where ultimate image quality is desired despite the source ... then you want to use the 10 bit 4:2:2 Uncompressed codec at the camera original resolution. ( 1440x1080 in the case of HDV. )

As far as the bit depth and such ... you aren't creating anything new- but by using a higher bit depth "container" you allow a tool like FCP to work in higher precision- so all the rounding errors in the 8bit 4:2:0 math end up in. That's good.

Now, you can apply the same thinking to a quality acquisition format like R3D or ArriRaw, but you have a whole different set of problems.

For one, once you reach that level of image acquisition, you will discover that the camera designers are very smart about being highly efficient with the data storage from the sensors they are working with. They have very high bit depths already, and those bit depths are already highly optimized in comparison to what the sensor can see.

Let me say that differently. With Red and Arri cameras the engineers optimize to the limit of the technology they have available. With HDV and DSLR they engineers optimize to the limit of the target price point.

So, basically don't screw with it. Work with the R3D directly (or the .ari files) until the end. If you must change formats, OpenEXR 16 bit is the one you want to use, and you probably also want to use ACES-IIF aware color control systems/applications.

If you don't know what that means ... then definitely don't screw with it.

Hi Alex, can you elaborate on this? With my former canon 5D workflow I was always taught to convert the H.264 to DNxHD 175, and then export to uncompressed, and make my deliverables from there. However are you saying that I was perhaps doing this all wrong? Or are you saying that there is a quality loss only when you convert footage into a higher resolution? If so, what would have been better with my workflow? I thought I had this figured out but now I wonder...

Thank you Alex!

 

[Alexander Ibrahim]

If all you are doing is straight cuts, this is true. But if you are working with effects, mixed footage, or hi-res plates, then the debate is whether it is better to take the upres hit at the beginning of the process while transcoding, or at the end ...

No, there really is no debate.

The general rule is to edit in the camera native source footage whenever practical. Only transcode on final output.

The reason to transcode DSLR h.264 footage at ingest is to increase performance in the NLE. H.264 footage is slow because of the processing burden on your system. Even GPU accelerated software gets quickly overwhelmed with multiple layers of h.264 footage on todays systems.

When transcoding you can avoid degrading the image. Its actually easy. Choose an uncompressed format that is equal or higher bit depth and equal or higher chroma subsampling. Technically, this is a transcoding step because the uncompressed format is technically encoded, but from a data integrity standpoint merely a decoding step from the source material.

So ... if I am working on a feature film shot with a 7D, I'd choose 8 bit Uncompressed 4:2:2 format. You could use 8 bit uncompressed 4:2:0, but I don't know any such codec. You could choose to convert to higher bit depth or better chroma subsampling ... but it doesn't really matter, you'd just be wasting bits - and unless you have a wicked fast system you'd just be limiting performance for no real gain.

Trust me 8 bit 4:2:2 uncompressed HD is bad enough on a budget - you do NOT want to borrow trouble.

The one exception is if I am working in Final Cut Pro. FCP defaults to working with 10 bit precision if you have a 10 bit timeline ... and the easiest way to ensure that all your editors (who you may not be supervising ... or even communicating with!) use this, is by making your footage 10 bits. So, for FCP post, I'll go ahead and bump the footage to 10 bit 422 Uncompressed. Note that this is not in order to solve a technological problem, but a human one. If you can control post, or train your editors properly, then 8 bit Uncompressed is plenty.

The data rate of uncompressed is its own limitation on performance. You can solve that bandwidth issue, but most likely, if you could spend cash to solve the bandwidth issue on HD footage ... well you could shoot a better camera.

I normally recommend just using ProRes 422 HQ (or occasionally DNxHD 175) because the project isn't being held to such high standards.

I prefer ProRes, because it defaults to being a 10 bit codec. DNxHD has a toggle switch for 8/10 bit ... so it introduces an extra detail to worry about. Also, in my side by side tests, I find that ProRes HQ holds up better to generational processing than DNxHD. I only choose DNxHD if I expect that I will have to interchange extensively with Windows systems.

(You should NOT be doing generational processing (i.e. multiple transcodes) ... but on projects where a 7D is the camera, you can't trust everyone in your post workflow to be competent.)

You absolutely are creating something new, you're creating the higher bit depth "container", and that is what makes all the difference in processing. And whenever you are applying an effect, you are creating new footage, literally in the case of Adobe. I would rather have that new content created at the higher resolution level; gives a better end result. Absolutely agree with all your comments regarding raw footage.

Wrong. Sorry.

You can set up any NLE to process timelines at higher bit depth, and that is all the real benefit you get.

In fact, NLE's are increasingly defaulting to 16/32 bit per channel floating point RGB processing. (Premiere 5.5 does this I believe, as does FCP X. Smoke has had this for a while.)

The only "new data" you may create comes from one sort of interpolation or another. Tools like 5DtoRGB (which I use to batch process 7D footage to ProResHQ) interpolate chroma data to create a fuller 422 data set - and that extra data is just made up. If you upscale resolution, its the same story. There is an emerging class of "smart" upscaling tools ... but they are just making stuff up using more complex (and hopefully better looking/more accurate) rules.

 

[Alexander Ibrahim]

Hi Alex, can you elaborate on this? With my former canon 5D workflow I was always taught to convert the H.264 to DNxHD 175, and then export to uncompressed, and make my deliverables from there. However are you saying that I was perhaps doing this all wrong? Or are you saying that there is a quality loss only when you convert footage into a higher resolution? If so, what would have been better with my workflow? I thought I had this figured out but now I wonder...

Thank you Alex!

Well, read my previous post and see if it helps.

It's a balancing act really.

I need to get out of the h.264 codec for performance reasons mostly ... but I don't want to degrade my footage too much or create other performance problems.

So, the best compromise I've seen is the use of the ProRes HQ or DNxHD 175 codec. Both are 10 bit formats with 422 chroma subsampling. Any conversion errors should result in new data that fits in the larger gamut colorspace ... in theory. The data rate (given in mbps, which comes out to roughly 22MB/s) is such that you can easily handle 2-3 streams on a single commodity SATA drive. (Not that I recommend that ...)

So, in my response to Randall Nott, I did mention that you do increase errors in transcoding to another format's chroma subsampling, just as well as transcoding to another resolution. That is exactly so.

If you have shot something on a crappy camera, like a DSLR, and you need it to go through complex post, then the best advice is to transcode that footage to uncompressed. If the post chain isn't going to be complex then I think you'll be very happy with ProRes HQ or DNxHD.

Since I am writing again ... I should mention ProRes 4444 and Cineform Filmscan formats.

Both are absolutely phenomenal formats. I kind of love them. Here's the thing, if you are transcoding "up" from DSLR h.264 they aren't better than intermediate formats like ProRes HQ. If you are using a tool that interpolates (resolution or chroma subsampling) you can get better looking results with those codecs - because the interpolation can take place on a finer scale. But is better looking really better?

Here is the trouble ... the only reason you might care is because you have complex post - for example, I may be keying and actor in front of a green screen, and I may have to consider the fine lines of hair (or lace), where you can see through it to the green screen.

What will "better" interpolation achieve in this circumstance? Well, its going to alter the actual fine lines that the camera imaged! This will almost always be a bad thing!

So, if you really need the quality to deliver a result ... you want to go with a straight un-interpolated decoding to an uncompressed format. (For VFX work, I recommend 16 bit OpenEXR, Uncompressed.)