2.35:1 Aspect Ratio & HD
Published : 9th August 2003
Next week, I will shoot some test for a Kung Fu film, located in Argentina. The idea is to blowup to 2.35:1
The technical path is as close as this
F-900 (Fujinon cinestyle 5-50)
Post (DeckLink HD in FCP)
I will appreciate any suggestions, tips, tricks and horror histories about the process.
My main concern regards an excessive loss of resolution with the less area used.
Thanks in advance
I don't have any horror story about that. It was always a good method. With my film recording and kinescoping experiences, HD 24p is a much more powerful than 35mm film(camera) as for 2.35:1 aspect ratio film-out.
Actually, it is not a blow-up process and there is no resolution loss. Film quality of 2.35:1 recording(or kinescoping) is much better than that of 1.85:1. If you test and look at it once, you'll be surprised. IT IS NICE! It is because 2.35:1 film frame size(super 35mm) is larger than 1.85:1 film frame size.
The only thing you have to do is to select 'VAR V' as 'aspect mode' at Operation Menu#3. It can be easier and exact if you use DCS CamAlign chart. If you don't have it, you may calculate VAR V number and put it on.
Adjust VAR V number - 1920:817=2.35:1.
I will also shoot a new Korean feature film with HD 24p in this May. I persuaded the director to use 2.35:1 aspect ratio (super 35) because it shows much better quality.
Also, 720P format is not suitable to 2.35:1 film-out, because a frame size becomes 1280x545 for 2.35:1. This number is not enough at all for a big screen, I believe.
HD DP in Korea.
I'm not sure what your specific concerns are but if you are considering framing for dual format output ie. 1.85 flat and 2.35 I can give you some of my experience.
A while back I shot something with the F900 that wasn't intended to be filmed out and was framed 16x9. Later the client found an alternative use for the footage and decided on a 1.85 filmout AND a 2.35 filmout. I was very nervous because the post path didn't allow for custom framings of the footage. The original 16x9 would be centre cropped horizontally for the 2 transfers to film without regard to head room.
I was pleasantly surprised at screenings at Technicolor of test film outs by how well the footage accepted this cropping and found it was not objectionable at all. Yes, I would have framed slightly differently if I had been shooting from the start for 2.35 but the way it was done did not turn into the disaster I expected.
From a quality/grain/resolution standpoint the 2 images looked almost identical to me, but then we did not see a side-by-side to know for sure.
One thing I found in projection at theatres that helped with the smaller HD image area to film neg space ratio was that in many cinemaplex theatres the screen width doesn't vary when changing from 1.85 prints to 2.35 prints. It seems most theatres use the same screen real estate width wise and vary the height of the picture.
One last thought. If you do shoot 2.35 and then the post path includes a 1.85 filmout later when the project is likely to be out of your control, you (everyone) will be screwed if you didn't keep a clean top and bottom frame while shooting. If it were me I'd keep mic booms and flags, etc. free of the between-1.85-and-2.35 zone while composing my shots.
Randy Miller, DP in LA
Koo, Jae-mo wrote :
>HD 24p is a much more powerful than 35mm film(camera) as for 2.35:1 >aspect ratio film-out.
Well, that's certainly an interesting observation. It didn't appear to be the case with Mr. Lucas' last epic, though. It looked pretty mushy, though usually the original photography was buried deep beneath a CGI stew.
> Actually, it is not a blow-up process and there is no resolution loss.
Well, it's not a "blow-up" -- technically -- because you are transferring from a digital medium to an analog one. But there is, for 2.35 (actually 2.4 these days) aspect ratio a definite resolution loss.
You are taking an image with a recorded resolution of about 800 pixels (when cropped to 2.40) x 1440 pixels (remember HDCam is filtered from 1920 to 1440 before recording) and showing it on a 40-foot wide screen. Divide those 40 feet (no metrification jokes!) by 1440 pixels and you'll find that each pixel will be 4" wide on the screen. (40 feet x 12 = 480 inches, 480/1440 = .333, or 1/3 foot, or 4 inches.)
A true 1920-pixel format (say Viper recorded to hard drives or D5 HD) would yield a 3" pixel.
Much of this depends on what sort of theatre you are going to be showing your film in. In the US, most theatres maintain a common top and bottom line for the image, and just open the curtains/masking to widen the image for 2.40:1 aspect ratio films.
If that's the case, (setting aside any issues of fitting 1.77:1 into 1.66 or 1.85), a screen that was 40 feet wide for 'scope would be masked to about 30 (rounded to keep things simple) feet. On this screen, a pixel from a camera that records 1440 pixels horizontally would be 3" wide, and a pixel from a camera that records 1920 pixels horizontally would be 2.25" wide.
Then, let's look at vertical resolution. For a 2.40:1 aspect ratio, you crop the 1080 pixel vertical resolution of HDCam to about 800 pixels. You are throwing away 280 lines of information -- nearly 25%. And showing this on a larger screen. Not a good thing. In this case, the Viper camera, which, through clever CCD design can output a full-resolution 2.40:1 aspect ratio image, would be a better choice -- no wasted pixels.
Of course, depending on the route the filmout takes -- direct to printing negative, or through an IP and DN to a print, and what stock is used for the filmout, and what film recorder is used -- all of these are variable that might mask the inherently low resolution of a 1440-pixel format.
>Film quality of 2.35:1 recording(or kinescoping) is much better than that >of 1.85:1. If you test and look at it once, you'll be surprised.
Yes, you do indeed use more film area of the negative here, but the resolution of the 35mm negative is so beyond what's recorded on HDCam that it's academic, IMHO.
I think there are more interesting ways to get a 2.4:1 image digitally, but that's a future discussion.
Jeff "still likes Academy aperture" Kreines
I've tested film-outs of HDCAM to 1.85 and cropped & stretched to 2.35 scope and the difference in resolution is minor, even with the loss of vertical resolution from 1080 to about 800 lines.
So if you like the 2.35 ratio, then go ahead. The one thing about 2.35 is that it is generally projected on a wider screen (not a shorter screen -- I've only seen that in some cheap multiplexes.) So the original digital image is being enlarged more. All this is to say that certain digital artefacts like edge enhancement become more obvious, so it becomes even more important to turn Detail off or way, way down. Noise problems also seem more obvious. But oddly enough, resolution seems similar to HDCAM-to-1.85. It may be that the 1.85 area of the print has to work harder to fill the screen. On the other hand, nothing is as variable in quality as scope projection in this country.
Cinematographer / L.A.
Jeff Kreines wrote :
>A true 1920-pixel format (say Viper recorded to hard drives or D5 HD) >would yield a 3" pixel.
Not quite. The Viper uses an oversampling CCD and line interpolation to allow it to record an anamorphic 2.35:1 image, so it yields a true 1920x1080 resolution on that image. This ability to do 'scope format without sacrificing resolution is one of the biggest advantages of the camera.
IATSE Local 600
Michael Most wrote :
>Not quite. The Viper uses an oversampling CCD and line interpolation >to allow it to record an anamorphic 2.35:1 image, so it yields a true >1920x1080 resolution on that image.
I clarify that later in my post -- the original mention of the Viper was just to make the 1920 vs. 1440 point -- I should have combined the two, but I was trying to go one step at a time.
The ability to do scope (2.37) format with the Viper without sacrificing resolution is a big advantage for the Viper camera, but the methodology is actually not as described recently on CML : "......
The Viper uses an oversampling CCD and line interpolation to allow it to record an anamorphic 2.35:1 image, so it yields a true 1920x1080 resolution on that image. This ability to do 'scope format without sacrificing resolution is one of the biggest advantages of the camera."
I checked with a Thomson Camera Development Engineer, as line interpolation is not part of the Dynamic Pixel Management process. His description follows:
Oversampling and interpolation suggests you are recalculating a pixel out of several others. This is indeed not the case. The sensor makes use of sub-pixels. For 2.37 you use another set of subpixels as for 1080 or 720. By using these subpixels, you can switch to a native format in the sensor. This is the DynamicPixelManagement principle.
GEORGE C. PALMER
HD and Digital Imaging Services
Hi all !
thank you all for your invaluable input, tips and experience.
If everything goes OK, we will shoot on Saturday at the roof of the National Library in Buenos Aires, I’ll let all of you know as soon as I have something to see, maybe I can post some stills.