Dragon....

[Eric Haase]

I have a "guess" on the usable range without noise reduction involved, and it'a good thing. But I want to hold off on it now because a couple of things.

1. Don't want to spread crazy info.

2. Want to really test it out myself before stating anything.

I will say this regarding noise. With Epic with Mysterium-X at 5K with a proper finish down to 4K at ISO ranging from 250-6400. Without noise reduction involved I can get a nice feel and look out of ISO 1280-2000. ISO 3200+ tends to be where I wrangle in some noise reduction personally. All of this really depends on what look you want out of your footage and how much texture you're okay with.

If you recall back to my older Epic and Scarlet MX noise tests you can see with a bit of noise reduction involved ISO 6400 and even usable ISO 12800 is possible. Just depends on how you work it really.

That said..... Knowing that Dragon's got more meat (dynamic range) in it's REDCODE RAW, lower noise floor, and general idea the advantages it will have in terms of latitude compared to Mysterium-X; just think what we could potentially do.

Wow those numbers are generous. It's really subjective but I draw the line on MX at 1000 and only go to 1280 if I absolutely have to. I'm only talking about these ISOs when the information in the scene is mostly below key, which is typically when you're using high ISOs. The range from zero to 20% on the MX can look pretty noisy even at 800 or 1000. This is why, like Probst, if the scene has most of the information in dark tones below key I tend to use as low an ISO as possible without clipping the highlights. In essence, get the fattest neg and print down. Also, youtube and vimeo tend to wash away noise with compression so things may look a lot less noisy through these media than on broadcast or projection.

 

[Matt Ryan]

Wow those numbers are generous. It's really subjective but I draw the line on MX at 1000 and only go to 1280 if I absolutely have to. I'm only talking about these ISOs when the information in the scene is mostly below key, which is typically when you're using high ISOs. The range from zero to 20% on the MX can look pretty noisy even at 800 or 1000. This is why, like Probst, if the scene has most of the information in dark tones below key I tend to use as low an ISO as possible without clipping the highlights. In essence, get the fattest neg and print down. Also, youtube and vimeo tend to wash away noise with compression so things may look a lot less noisy through these media than on broadcast or projection.

I agree. I rarely have material I'd use at anything above 800ISO on MX even at 1080p, and no way in hell at 6400 or 12800. I do not like noise... I'm really excited if Dragon truly is cleaner at 2000ISO than MX at 800. Imagine how useful that will be and how huge that is to also have 16+ stops of DR.

I also hope tungsten performance is improved and that we get noise free images at 800ISO+ without any blue noise creeping in.

 

[Phil Holland]

Wow those numbers are generous. It's really subjective but I draw the line on MX at 1000 and only go to 1280 if I absolutely have to.

Well yes. And again it depends on what you are doing and how much "noise texture" you can live with. I should be clear here that the vast majority of my shooting lands in the 320-800 range (a good chunk of what I shoot ends up as a VFX element or plate of some sort) favoring lower ISO when possible.

I don't work for youtube or vimeo low bit rates...... I tend to deliver 16bit TIFF, EXR, or DPX to studios.

Here's a very recent 5K 2:1 ISO 1280 (Panavision Primo 35mm T1.9).




The sweet spot on Mysterium-X is indeed in that 320-800 range for sure if you're after a clean image.

 

[Joel Arvidsson]

Are we going to se you lighting up a cigar in the basement Phil=)

 

[Karim D. Ghantous]

The problem with skin tones seems to come from the attitude that colour temperature can always be corrected post-exposure. That is how I interpret Timur Civan's blog post:

http://timurcivan.blogspot.com.au/2013/06/extracting-best-color-from-your-red.html?m=1

If you want to preserve ambient colour temperature, no problem, as I understand it. But if you think you can correct anything in post, you probably don't fully appreciate the technology that you're using, be it film or digital. All IMHO.

 

[Phil Holland]

Back to Dragon.

I have spent a good amount of time thinking about Dragon since it's initial announcement and have been fortunate enough to get an early look at the camera a few times.

Something has been on my mind. Aspect Ratios and Production Workflow (and a bit of historical relevance for that matter).

I work primarily in features and our world is more or less a 2.40:1 (2.35:1, 2.37:1, 2.39:1 - your choice really) and 1.85:1. UHD broadcast content will exist mainly as a 16x9 (1.78:1) format.

With my film background in 16mm, 35mm, VV, and 65mm combined with Dragon's native sensor format landing at a nice and big 1.94:1 with a resolution of 6144x3160 pixels I have thought about how to use this beast in a variety of workflows.

Then Jarred let the cat out of the bag on the customization of user created resolutions and formats and I really got to thinking. About everything.


So here's the pitch and I'd love to hear your thoughts on the matter really.

I would like to see a 6K format of 2.28:1 6144x2700 adopted by productions and possibly a new format within the Epic Dragon itself. Why? Let me explain.




Key Advantages and Features of Proposed New Format:
- 6K WS and the various variations like 2.40, 2.39, 2.37, 2.35 can be extracted from the frame. (with a bit of room to re-frame even)
- All Official Red 5K Formats can be extracted from the frame.
- Proper 5K to 4K debayer/resize on 5K formats can be extracted on projects shot in the new wide format even when cropped to other UHD formats.
- Data rate savings by shaving off 6144x460 pixels, perhaps a bit higher REDCODE compression ratio and/or higher frame rate.


I feel that if 2:1 (which is one of my favorite formats straight out of the camera) exists, which is just a hair under the max vertical resolution of the sensor, then this new 2.28:1 format could exist being a hair and a half over the 2.37:1 resolution. Or perhaps even become what we would call 6K WS, or more likely, a brand new format name. Personally I'd go for "6K WV", "6K VS", "6K WM" as to me it's a WideVario or VarioScope format.

For "epic" cinematic experiences I can see this a very powerful tool in terms of mastering to 4K 16x9 with and without "bars".

The 2.28:1 aspect ratio itself does indeed have some actual format relevance primarily to 65mm cinematography as it is indeed the aspect ratio of 65mm 5-perf film/Showscan (not the same negative size), which some of the film folks out there still love as a format. Maybe even call it "WM SS" for Doug


Makes a lot of sense to me. Does it make sense to you guys? Red guys? Am I crazy? It is 3:15am. I could be crazy.

 

[Karim D. Ghantous]

Oh, and can someone tell Jim that the Nikon D800 doesn't shoot at 10fps but at 4fps. He closed that 'Always learning' thread. And also tell him that DxO scores are off. Many people prefer the Leica M9 to a lot of DSLRs, yet its score is lower. Go figure.

 

[Eric Z]

Interesting proposition, Phil.

I like 2:1 (6144x3072), nice and even.
Easy to manipulate in post to reduce to other formats, easy to remove camera shake, easy to remove vignetting.

Plus, Jim seems to favor it as well

 

[Häakon]

Hey Phil,

Interesting proposition! I don't think the cinema industry cares what I think about aspect ratio standards, so I'll leave that to someone else. I am interested however to know if there are 16:9 and 1.85:1 ratios planned that use the full height of the sensor (or very close). The way your diagram shows, there's massive loss of sensor real estate when framing for either of those ratios which not only means reduced resolution but a greatly narrowed FOV as well. I'm gathering that the intent is to use your newer ratio as the standard and "protect" for everything else, but for situations where it's known that the footage is going to be used in a broadcast space this doesn't make much sense. Perhaps you just left those two choices off the diagram as to focus soley on cinematic use, but I just wanted to clarify.

 

[Phil Holland]

Hey Phil,

Interesting proposition! I don't think the cinema industry cares what I think about aspect ratio standards, so I'll leave that to someone else. I am interested however to know if there are 16:9 and 1.85:1 ratios planned that use the full height of the sensor (or very close). The way your diagram shows, there's massive loss of sensor real estate when framing for either of those ratios which not only means reduced resolution but a greatly narrowed FOV as well. I'm gathering that the intent is to use your newer ratio as the standard and "protect" for everything else, but for situations where it's known that the footage is going to be used in a broadcast space this doesn't make much sense. Perhaps you just left those two choices off the diagram as to focus soley on cinematic use, but I just wanted to clarify.

Well yes, for cinematic use and UHD use.

I believe a 6K HD format (or very close to it) will exist in terms of reaching the top of the sensor. There will be of course a 6K Full format and the ability to set your own resolution/aspect ratio (which I think is still planned). My primary reason of coming up with this idea was motion pictures that prefer wide screen format capture and wish also have the flexibility to have a proper UHD 16x9 finish within the same format itself.

So for instance, The Hobbit lands at 2.35:1, but if captured and worked on in this particular format a true 4K debayer 16x9 image could be cropped out of if to maintain UHD quality levels within that new 6K format. Or they could just simply scale down the widescreen image to have black bars above and below the image on 16x9 screens.

Additionally for those working for that WS format the weird variants of 2.40, 2.39, 2.37, and 2.35 all become "vertical" only crops. I see an advantage to that.

More or less I'm thinking of the 4K master and how we get there.

It's just a thought really. Something that I think could be very flexible as a WS format.

 

[Martin Stevens]

I could be crazy.

No, you are not crazy, and your sensor cropping ideas look fine to me.

I am assuming your new aspect ratio works well with all S35 image circle cine lenses?

 

[Alain M]

Beautiful Still from pulled from your footage Phil!! Share more when you can...

Alain

 

[Brant Hadfield]

Will Dragon's Highlight Roll-off characteristics be a function of its dynamic range, new OLPF, or color science or curve? Will a smoother roll off than MX now be possible by adjusting curve, because Dragon has so much DR to start with?

 

[Mark Toia]

If Dragon performs anything like the monochrome did in these tests,... I'd be more than impressed. (fingers crossed)
When we shot these tests, I couldn't see hardly anything with my own eyes, it was pitch black in some of these shots, Only when I looked through the view finder was there actually any detail.
It was like looking though a night vision camera.
We were shooting between 2000 & 4000asa... 4000asa was still very useable.

 

[Martin Stevens]

Awesome as always Mark!

 

[Alain M]

WOW!!! Thanks Mark... it looks awesome

 

[Kyle Stryker]

Back to Dragon.

I have spent a good amount of time thinking about Dragon since it's initial announcement and have been fortunate enough to get an early look at the camera a few times.

Something has been on my mind. Aspect Ratios and Production Workflow (and a bit of historical relevance for that matter).

I work primarily in features and our world is more or less a 2.40:1 (2.35:1, 2.37:1, 2.39:1 - your choice really) and 1.85:1. UHD broadcast content will exist mainly as a 16x9 (1.78:1) format.

With my film background in 16mm, 35mm, VV, and 65mm combined with Dragon's native sensor format landing at a nice and big 1.94:1 with a resolution of 6144x3160 pixels I have thought about how to use this beast in a variety of workflows.

Then Jarred let the cat out of the bag on the customization of user created resolutions and formats and I really got to thinking. About everything.


So here's the pitch and I'd love to hear your thoughts on the matter really.

I would like to see a 6K format of 2.28:1 6144x2700 adopted by productions and possibly a new format within the Epic Dragon itself. Why? Let me explain.




Key Advantages and Features of Proposed New Format:
- 6K WS and the various variations like 2.40, 2.39, 2.37, 2.35 can be extracted from the frame. (with a bit of room to re-frame even)
- All Official Red 5K Formats can be extracted from the frame.
- Proper 5K to 4K debayer/resize on 5K formats can be extracted on projects shot in the new wide format even when cropped to other UHD formats.
- Data rate savings by shaving off 6144x460 pixels, perhaps a bit higher REDCODE compression ratio and/or higher frame rate.


I feel that if 2:1 (which is one of my favorite formats straight out of the camera) exists, which is just a hair under the max vertical resolution of the sensor, then this new 2.28:1 format could exist being a hair and a half over the 2.37:1 resolution. Or perhaps even become what we would call 6K WS, or more likely, a brand new format name. Personally I'd go for "6K WV", "6K VS", "6K WM" as to me it's a WideVario or VarioScope format.

For "epic" cinematic experiences I can see this a very powerful tool in terms of mastering to 4K 16x9 with and without "bars".

The 2.28:1 aspect ratio itself does indeed have some actual format relevance primarily to 65mm cinematography as it is indeed the aspect ratio of 65mm 5-perf film/Showscan (not the same negative size), which some of the film folks out there still love as a format. Maybe even call it "WM SS" for Doug


Makes a lot of sense to me. Does it make sense to you guys? Red guys? Am I crazy? It is 3:15am. I could be crazy.

I think this is a really great idea.

 

[Phil Holland]

No, you are not crazy, and your sensor cropping ideas look fine to me.

I am assuming your new aspect ratio works well with all S35 image circle cine lenses?

Well 5K technically is the proper S35 image circle on Dragon. 5.5K is essentially the same size of Mysterium-X.

This 2.28:1 format would be ideal for those shooting 6K "cinematic wide screen motion pictures" while maintaining sort of a mathematically correct debayer logic and also having 2700 pixels of height to extract any "less wide" formats within the 2.28:1 frame. It has certain advantages of 6K WS in terms of having a "vertical only crop" for WS and "horizontal only" crop for things like 5K, 5K 2:1, 5K HD, or say 1.85:1 out of the frame itself.

I need to be clear here, I'm thinking about this from a production workflow concept for those who really want to utilize the full width of the 6K Dragon sensor and see it through post/deliver as well. There will be many folks just shooting at 6K and 6K 2:1, myself included. However, I think this format is smart for feature production in a lot of ways. Especially considering where most of the new 4K displays land aspect ratio-wise.

I think this is a really great idea.

Thanks Kyle. I had this on my mind nearly a year ago since about August or September 2012. The more I thought about it, the more it makes sense as a format. Now that we're getting close to Dragon taking off I'd like to hear other's opinions on the concept.

 

[Martin Stevens]

Okay, so Phil, on the movie "The Girl with the Dragon Tattoo" they shot 5K for 4K, and from my understanding
they did not resize the image in post, they simple extracted the 4K right out of the 5K footage because
they thought it looked better etc.

So, do you prefer the "slight" resize from 5K to 4K for a 4K finish, or do you think a straight
4K extraction looks better?

 

[Phil Holland]

So, do you prefer the "slight" resize from 5K to 4K for a 4K finish, or do you think a straight
4K extraction looks better?

Heh. I think it's up to the shooter/director/whowever to decide what looks better on a project by project basis. Luckily, we have the freedom within our Red cameras to do this.

For productions shooting 5K today we are scaling down to our 4096x2160 (or 3840x2160p) frame size, which in theory provides the . Many shoot straight up 4K too. Just depends on the look you're after.

Basically I'm keeping in mind the Nyquist-Shannon theorem and applying that concept to the new higher resolution Epic Dragon sensor while focusing on the relevance of certain standard formats along the way.

In terms of resolution and freedom 6K Full is the good sauce, but may not be totally necessary on a project by project basis. Just like if you were shooting for say.... 6K HD for a UHD deliver. Just depends.

The viability of this format is that you can crop or resize for a variety of motion picture standard formats within it "if" you plan on creating a wide screen production from the get go. This particular format makes a bit more sense to me from an acquisition and post standpoint for productions capturing in a WS format.

Simplified, it's a lot like 6K WS except with more headroom. That should benefit WS and productions that "may" crop to say 16x9 UHD displays if they want to cover that full screen and still have stunning imagery.

Additionally and as a total side not between the pixel design and format size there's a lot of talk about Dragon 6K matching or exceeding the visual quality of 65mm 5-perf, which is indeed this particular aspect ratio. It's a format that's still actually used today on larger budget productions that shoot 65mm. So in some ways a bit of a nod and seductress for folks who enjoy that format.

That's pretty much why 2.28:1 makes a lot of sense to me.