Still confused about lens cropping factors . . .

[Ralph Oshiro]

Okay, I'm looking for single-figure answers here. I appreciate all the format tutorials and informative diagrams, but I still don't have confirmed answer on this question . . .

Acquiring a 1080p, 1.78:1-aspect ratio (16:9) image using RED, what is the lens focal length multiplication factor [cropping factor] when using Nikon 35mm still lenses? For example, if using a 14mm Nikkor still lens on a RED camera, what "135-format equivalent" lens would it look like on the RED camera in millimeters? Similarly, what are the following values for 'x' using any given value for 'y'?

1. Given 1080p-resolution RED acquisition in 1.78:1 aspect ratio:
135-format equivalent in mm = x cropping factor x 135-format lens of y mm

2. Given 2K-resolution RED acquisition in 2.35:1 aspect ratio:
135-format equivalent in mm = x cropping factor x 135-format lens of y mm

3. Given 1080p-resolution RED acquisition in 1.78:1 aspect ratio:
135-format equivalent in mm = x cropping factor x 2/3" B4-format lens of y mm

4. Given 2K-resolution RED acquisition in 2.35:1 aspect ratio:
135-format equivalent in mm = x cropping factor x 2/3" B4-format lens of y mm

My brain hurt just trying to ask this question correctly! Again, I appreciate all the tables and reticle diagrams here, but if someone could please create a table that also includes the column, "focal length cropping factor" (or whatever you want to call it), I would be greatly appreciative.

[Martin Drew]

Ralph.

Firstly, the multipliers discussed relate to horizontal FOV, which will be the same for 1.78:1 and 2.35:1 (assuming you are deriving 2.35:1 by cropping the 1.78:1 format).

I am assuming that when you talk about acquiring in 1080p you mean acquiring using a 1080p window on the sensor, which would normally be done using a 2/3" format lens. The relevant point is that you are windowing the sensor rather than downrezzing. If that is the case then:

1. x = 3.46154 x y
So if you use a 14mm lens the image would be framed the same as if you had shot with a 48.5mm lens on 135 (35mm stills) format

2. x = 3.24324 x y
So if you use a 14mm lens the image would be framed the same as if you had shot with a 45.5mm lens on 135 (35mm stills) format

3. x = 1 x y
So if you use a 14mm lens the image would be framed the same as if you had shot with a 14mm lens on 2/3" (B4) format

4. x = 1.06731 x y
So if you use a 14mm lens the image would be framed the same as if you had shot with a 15mm lens on 2k (s16) format

I hope I have understood your query correctly. If not then let me know because it is quite easy getting tied up in knots with this stuff.

Martin

[Ralph Oshiro]

Martin:

Thank you for answering my questions! Well, I guess that's not quite what I wanted to hear (I mean, it sounds like you answered them correctly, those just aren't the answers I was hoping for). Must I "window" the sensor when shooting <4k resolution? I had originally assumed that everything was "seen" through the entire sensor at 4k, then downrezzed in-camera to derive the other resolutions, so that there was no large multiplication factor. This isn't so, right? But while reading a related thread, "windowing," isn't just seeing the center part of the sensor (which would not explain your high multiplication factors). So, the only way to shoot with RED, with a minimum multiplication factor when using 135-format still lenses, is to shoot at fulll 4k resolution? Thanks in advance for your reply..

[Martin Drew]

Okay. Now I understand you better, you are talking about scaling in camera. You can acquire at 4k and then scale in camera to 1080p or 720p. As soon as you scale in camera you lose the RAW format so then you are recording in Recode RGB which doesn't compress quite as well and isn't a RAW format (so you can't take it into RedCine). For all these options the multiplication factor for the FOV will be the same:

x = 1.62162 x y

That is using 4k of the sensor. This isn't actually the full sensor size, the full sensor is 4.5k, but the only way of recording that at the moment is uncompressed and unscaled to an external (very) high speed raid.

The other option is to record 4k redcode RAW and then scale to your final format, 2k or 1080p in RedCine. This makes a lot of sense because the data rate for 4k 24 fps Redcode RAW isn't that much higher then the rate for 1080p 24 fps Redcode RGB, especially if you take into account the added flexibility that shooting in RAW brings you. Data rate predictions made by Mike Curtis (and bear in mind the 1080p is an educated guess based on the 4k rate) are:

1080p 24 fps Redcode RGB = 18.1 MB/sec
4k 24 fps Redcode RAW = 27.5 MB/sec

http://web.mac.com/mikedcurtis/iWeb/...codeRates.html

If you want to get 2k directly in camera, this can't be scaled from 4k you have to acquire at 2k and then you will use the factor

x = 3.24324 x y

Of course Acquiring directly in 2k or 1080p windowed makes much more sense if you are using a S16 or 2/3' lens.

Martin

[Stuart English]

Another way of thinking about this is as follows -

4K capture on RED has the same horizontal width as Academy 35 mm.

1080p is created from that (by scale and crop) in REDCINE.

4K = 4,096 pixels wide 1080p = 1,920 pixels wide.

So if you do a 2:1 scaling in REDCINE (which is the fastest scale) 4,096 -> 2,048 and then crop that to 1920 pixels.

For full 4K image area you could scale 4,096 to 1,920 which is a scaling factor of 2.133 - You need more filter taps so it'll be slower.

[Nick Shaw]

…Recode RGB which doesn't compress quite as well and isn't a RAW format (so you can't take it into RedCine).

Is that correct? I was under the impression that REDCODE RGB could still be recorded un-processed, and that it could still be passed through REDCINE to adjust gamma, curves, etc.

Stuart, is this option still available, and if it is can REDCODE RGB be recorded 12-bit linear to maximise flexibility when doing this?

Nick

[Martin Drew]

Ooops Sorry Nick. You're right. It can still go through RedCine, just in a more compromised format.

M

[Ralph Oshiro]

Okay, I think I'm finally getting it--thank you to all who contributed their pearls of wisdom here . . .

Quoted from Stuart English's post in the, 'How many windows' thread . . .

"Two windowed sensor areas are currently allowed—
4K - Academy 25 mm
2K - Super 16 mm / B4 using adaptor
4.5K (not a window of course) - Super 35 mm

Note 1080p and 720p recordings are designed to have the same FOV, due to the design of the internal scaling engine. This means that 1080p, 1080i and 720p footage will all have the the same optical characteristics."

So as I now understand it (or mis-understand it), here's my take:

1. Acquiring in 2K-windowed has no benefit to me if my depth-of-field characterisitics utilizing a "2K window" are only slightly more shallow than those resulting from using a B4 lens on a standard 2/3" camera (or identical, in the case of lensing with 2/3" B4 lenses, and scaling in-camera from the 2K window to a 2/3" image area).

2. Working with a 3x-plus multiplication factor is impractical for use with standard 135-format still lenses, if you want to utilize very wide-angle optics.

3. Acquiring in 4K RedCode RAW gives me both the depth-of-field characterisitics (my primary motivation for purchasing RED) that I associate with 35mm motion picture cameras, and the most easily optimized datastream for post-processing in RedCine. Clearly, this seems to be the only desirable acquisition resolution to achieve my desired S35mm depth-of-field characterisitics. But without performing any in-camera scaling, I have a much larger data storage requirement, which exceeds my near-term delivery requirements (my current deliverable would only be a 2.35:1-aspect ratio, 1080p data image).

4. If acquiring in 4K RedCode RAW, while performing no in-camera scaling, how many minutes of 24p video can be stored on the on-board 160GB RED-DRIVE or on a 128GB RED-FLASH media magazine?

[EDIT: Quoted from jbeale in the 'Data Rate and Storage Options' thread, "Last I heard, 24 fps 4k REDCODE is about 27.5 MBytes/s or 220 megabits per second, so a 320 GB Raid array (2x160 GB) attached to the RED could hold just over 3 hours of 24 fps, 4k material." So, I guess the answer is about an hour or more for a 160GB RED-DRIVE]

5. Acquiring in 4K RGB, and scaling in-camera to 2K, still gives me the depth-of-field characterisitics that I associate with 35mm motion picture cameras, and a slightly reduced data storage requirement, but at the expense of a less desirable datastream for post-processing in RedCine. Rescaling in RedCine, at anything other than an exact 2:1 scaling ratio, would also incur significantly greater (possibly impractical)rendering times.

Is this all sort of correct?

[Ralph Oshiro]

So . . . my workflow will be:

1. Acquire in RedCode 4K RAW to on-board 160GB RED-DRIVE w/ 60-90 minute capacity.

2. Rescale and crop to 2:35-aspect ratio, 1080p in RedCine.

[Rob Lohman]

Is that correct? I was under the impression that REDCODE RGB could still be recorded un-processed, and that it could still be passed through REDCINE to adjust gamma, curves, etc.

Stuart, is this option still available, and if it is can REDCODE RGB be recorded 12-bit linear to maximise flexibility when doing this?

Nick

No, you cannot load REDCODE RGB back into REDCINE. There's really no need either.