F55: PMW-F55 Sensor Details

T

Tom Roper

Veteran
Note: This is PMW-F55, not PXW-Fs5

Background:

The F55 sensor from the best information I can find is:

24mm x 12.7mm
17:9 aspect ratio
11.6 mp total
8.9 mp effective pixels

The 8.9mp "effective pixels" is said to be a 4k-DCI crop of the total pixels, 11.6mp.

I've done the math, if you want to participate you do the math.

Bottom line:

If true the 8.9 is a crop, then it's a 1.82 crop of the full frame (36mm x 24mm) image circle. I don't believe the crop factor can be that much. That's more than half way to micro 4/3.
On the other hand, if the sensor is over sampled 4k DCI, the crop factor is in line with other S-35 sensors from Red, Canon, Arri.
 
FWIW, there are a few full frame cameras that have ~1.8x 4K crop (mostly from Canon and Nikon).
 
NorBro said:
FWIW, there are a few full frame cameras that have ~1.8x 4K crop (mostly from Canon and Nikon).
Click to expand...

Not sure which cameras exactly have the 1.8 4k crop, but I ran some numbers anyway for R5 and R6.

R5: 4k UHD crop at 1:1 pixel = FF crop factor 2.29
R6: 4k UHD crop at 1:1 pixel = FF crop factor 1.55
 
So the R5 and R6 numbers even though it looks okay for R6, are not because instead of 1:1 pixel crop is based on 5.1k or 5.1k 94% oversample.
 
The 5D Mark IV and EOS R from Canon. I'd have to find the Nikon.

But I'm not sure what you mean about the R5...there is no crop in FF? (But there's a separate S35 mode.)
 
NorBro said:
The 5D Mark IV and EOS R from Canon. I'd have to find the Nikon.

But I'm not sure what you mean about the R5...there is no crop in FF? (But there's a separate S35 mode.)
Click to expand...

I wasn't sure, was only figuring it as if it did have a crop. Since it doesn't, just ignore it. And since the R6 uses 94% of a 5.1k sensor area, it would have to be recalculated as well.

I'm looking to understand the discrepancy of the F55 in the first post. You had said some FF cams had a 1.8 4k crop. If they did, it should not be a "S-35" crop.
 
It's not...it's a 1.8x crop or around there for the two mentioned above (and a few others, but it's rare).

Then other cameras are full-frame (or very close) with S35 crop modes (or very close).

Anything around 1x is considered FF, 1.5-1.6x S35, and anything above that is when people start talking about the camera having a "crop" (which is usually a bad thing like with the 5D Mark IV which caused major disappointment).

With MFT it's of course expected.

Sony made millions (I presume) by offering FF 4K (with other limited specs) very early in this game.
 
Tom Roper said:
Bottom line:

If true the 8.9 is a crop, then it's a 1.82 crop of the full frame (36mm x 24mm) image circle. I don't believe the crop factor can be that much. That's more than half way to micro 4/3.
On the other hand, if the sensor is over sampled 4k DCI, the crop factor is in line with other S-35 sensors from Red, Canon, Arri.
Click to expand...

Well the math took me rather longer than I'd care to admit, but if the 8.9m pixels is a crop of the 11.6m, and the 11.6m are located within the 24x12.7mm surface area, then the crop factor would be approx 1.71x (not 1.82x).

The question is whether 24 x 12.7mm is the total or active area of the sensor?

I might dig a little deeper into this.
 
Grug said:
Well the math took me rather longer than I'd care to admit, but if the 8.9m pixels is a crop of the 11.6m, and the 11.6m are located within the 24x12.7mm surface area, then the crop factor would be approx 1.71x (not 1.82x).
Click to expand...

1.82x was based on 11.6 mp over 24x12.7mm sensor area, (calcs below) but you can forget it, the answer is 1.6x crop factor over 24x12.7mm sensor area for 8.9 mp.

Assume:

FF sensor 36mm x 24mm
Camera sensor 24mm x 12.7mm
11.6mp
Aspect Ratio 17:9

Let CF = 35mm FF crop factor
Let PP = pixel pitch mm
Let Icf = image circle of FF sensor mm
Let Icc = image circle of sensor crop mm

CF = Icf/Icc
PP = 12.7/(SQR(11.6e+6 *9/17))
Icf = SQR(36^2 + 24^2)
Icc = SQR((PP*4096)^2 + (PP*2160)^2)

PP = 5.125e-3 mm
Icf = 43.267 mm
Icc = 23.732 mm
CF = 1.823
 
Hmm... I got the following (on the assumption that the 11.6m is contained within the listed 24x12.7mm sensor area):

Sensor
Ratio 17:9
Size 24mm x 12.7mm
Total Pixels 11,600,000
Effective Pixels: 8,900,000

Larger Area
24mm x 12.7mm = 304.8mm2
304.8 / 11,600,000 = 38,057.7 pixels per square millimetre

Length per millimetre = √38,057.7 = 195.08px per millimetre

Pixel Pitch
1mm / 195.08px = 0.0051mm per pixel = 5.1 microns per pixel

Smaller Area
8.9m/11.6m = 0.767
304.8mm2 x 0.767 = 233.86mm2

4k Crop Dimensions
4096 / 195.08px = 20.99mm
2160 / 195.08px = 11.07mm
20.99mm x 11.07mm = 232mm2 (the 1.86mm2 difference here we can put down to rounding errors in this case)

Crop Factor
36mm / 21mm = 1.71x
 
Crop factor is ratio of imaging sensor diagonals. You used just the horizontal. The diagonal is the diameter of the image circle.

Example:

36 x 24mm FF sensor
24 x 12.7mm Sony sensor

If you only consider the horizontal, you get a CF = 1.5 (36/24)
But if you calculate the hypotenuse, CF = 1.593 (43.267/ 27.153)

That percentage, 1.593/1.5 times your 1.71 gives my 1.82.

Redo your calculation of the 21 x 11mm crop area using the ratio of diagonals, and you will get the same number as mine.
 
Tom Roper said:
Crop factor is ratio of imaging sensor diagonals. You used just the horizontal. The diagonal is the diameter of the image circle.

Example:

36 x 24mm FF sensor
24 x 12.7mm Sony sensor

If you only consider the horizontal, you get a CF = 1.5 (36/24)
But if you calculate the hypotenuse, CF = 1.593 (43.267/ 27.153)

That percentage, 1.593/1.5 times your 1.71 gives my 1.82.

Redo your calculation of the 21 x 11mm crop area using the ratio of diagonals, and you will get the same number as mine.
Click to expand...

Yeah, that’s the problem, you’re comparing the diagonals of a 3:2 aspect ratio to a 17:9 aspect ratio - which isn’t accurate because you wouldn’t be shooting in a 3:2 ratio (or if you were, you have to crop the sides of the smaller 17:9 format to match, and the crop factor would be even larger then).

We use the horizontal to measure crop factor because we shoot in “landscape” oriented frames that are always shorter than they are wide, and it’s therefore always going to be the width of the sensor that determines the FoV we’ll get from a particular focal length of lens.
 
Still, the 21 x 11 rectangular crop you correctly figured, it was only the CF number that needed adjustment. What remains unanswered is whether the 11.6 mp is contained within the 24 x 12.7 physical area for a 1.82 CF and 21 x 11 8.9 mp effective pixel area, or whether the true physical size of the sensor is larger than 24 x 12.7, the effective area containing 8.9 mp for a 1.59 CF. And then the elephant in the room, what if the 24 x 12.7 area was binned or down sampled from the 11.6 mp? Alister and others have said no, but where is the proof?

The only tools I have left to try may be similarly uncertain. That would be to put the F55 side by side with the P6k, and fit the same focal length prime on each and compare for a FOV difference. I can't fit the same lens on each because of the mounts but I can put the same focal length on each. The sensor in the P6k is said to be a little undersized as well but it has been variously reported to have a CF 1.55-1.60. If the CF 1.82 is correct for the F55 it would seem it should have a narrower FOV.

I appreciate that you have given your time to participate, also appreciate the CF calculator; nice to have a fast check, these calcs get tedious doing them manually. Back in the day I would take my T.I. programmable scientific calculator to class but for exams we were only allowed pencil and paper.
 
Grug said:
We use the horizontal to measure crop factor because we shoot in “landscape” oriented frames that are always shorter than they are wide, and it’s therefore always going to be the width of the sensor that determines the FoV we’ll get from a particular focal length of lens.
Click to expand...

The angle of view can be the horizontal, vertical, or diagonal. Take your pick. There is no rule that you must take the longest side of your frame. Anyway, the longest would be the diagonal. The diagonal is useful because it is the same thing as the diameter of a circle that outlines the rectangle. It is for the very reason that you might deal with rectangles of different proportions that I think it is useful to compare by the diagonal, as a kind of average of width and height.

Also, the convention is to use the diagonal: "The most commonly used definition of crop factor is the ratio of a 35 mm frame's diagonal (43.3 mm) to the diagonal of the image sensor in question" --- https://en.wikipedia.org/wiki/Crop_factor
 
combatentropy said:
The angle of view can be the horizontal, vertical, or diagonal. Take your pick. There is no rule that you must take the longest side of your frame. Anyway, the longest would be the diagonal. The diagonal is useful because it is the same thing as the diameter of a circle that outlines the rectangle. It is for the very reason that you might deal with rectangles of different proportions that I think it is useful to compare by the diagonal, as a kind of average of width and height.

Also, the convention is to use the diagonal: "The most commonly used definition of crop factor is the ratio of a 35 mm frame's diagonal (43.3 mm) to the diagonal of the image sensor in question" --- https://en.wikipedia.org/wiki/Crop_factor
Click to expand...

Mathematically? Certainly, you can calculate a number however you like. But for the purposes of video/motion-picture production? Crop factor absolutely has to be determined by the horizontal - because (in all but the most-obscure of circumstances) the horizontal axis is always going to be the longest, and because the entire purpose of calculating a crop-factor for motion-picture work, is to understand what the horizontal field of view for a given lens on a given format is going to provide (when compared to the 36mm wide frame of a 135/full-frame aperture).

The diagonal can only provide an accurate measurement for this if both sensors being compared, have identical aspect ratios.
 
Grug said:
the entire purpose of calculating a crop-factor for motion-picture work, is to understand what the horizontal field of view
Click to expand...

On what principle? I dispute the premise.

The diagonal can only provide an accurate measurement for this if both sensors being compared, have identical aspect ratios.
Click to expand...

I would say the opposite: the horizontal can provide an accurate measurement only if both sensors have identical aspect ratios. In fact, I did say the opposite, in my last post. You're simply contradicting me without any supporting explanation.

It's impossible to get identical compositions from frames of different shapes, as you know. So the best thing is to go for a similar overall feel. And the best overall feel is approximated by the diagonal. As an extreme example, a 50mm lens on a 36x24mm frame will feel very different than on a 36x9mm frame, though they have the same horizontal angle of view. A 43mm lens (if you can find one) would better match the overall feel, since it exactly matches the diagonal angle of view. You're unlikely to have a 4:1 aspect ratio as with my 36x9mm example, but the 2.4:1 aspect ratio that some people go for is different enough from still photography's 3:2 that merely matching horizontal will give you a tighter-feeling shot. Even so, some people will think we're being pedantic, because to match horizontal you would use a 50 on both and to match diagonal you would use a 45.
 
You must log in or register to reply here.
Back
Top