androidlad

4K 24-60P and 4K 120P have almost the same level of sharpness and detail. And 1080P is not supersampled, but pixel-binned from 24-60P and line-skipped from 100-240P.

Someone dissembled R5 and found barely any heatsink and shockingly there was no TIM (thermal paste).

Regarding the so-called "16bit linear RAW" 4264 x 2408 60FPS HDMI output, it's actually derived from 12bit ADC sensor readout, then dithered(upscaled) by the DSP to 16bit.

22 minutes ago, gt3rs said:

On the comments he says that was 4k 30 on both but he does not recall if HQ or not....

It was shot in 4K normal.

R5 8K/4K HQ: 21ms

R5 4K normal 24P to 120P: 8ms

A7S III 4K 24P to 60P: 15ms

A7S III 4K 120P: 8ms

R5 has better rolling shutter than A7S III:

The white dot next to the AF assist lamp is actually an XYZ sensor (visible light + IR sensor) to help improve AWB accuracy.

39 minutes ago, Django said:

Hmm.. so now he's actually advising people to ignore the overheating tests. The plot thickens..

Context:

This was tested in Florida under direct outdoor sunlight with ambient temperature of over 38C.

A7S III lasted 23min shooting 4K60

R5 lasted 34min shooting 4K30 HQ.

None of the 1080p modes on A7S III are supersampled from full sensor readout, 24p to 60p are pixel binned, 120p is pixel binned and lineskipped, 240p is futher pixel binned lineskipped (to 1420 x 804)

XAVC R5, not RS.

It's not a codec. The camera records XAVC full flavour internally and output X-OCN RAW to the Sony R5 recorder

It's called XAVC RS.

Shoot HLG, expose conservatively and push the shadow/midtone in grading. HLG has great latitude and BW helps mask the slight chroma noise.

The 2x2 pixel-binning in 4K normal and hfr mode has a softening effect, but it can be sharpened to be almost identical to 4K HQ (Oversampled from 8K).

3 minutes ago, Jay60p said:

Then would you agree that in general a 12MP camera (with more moire/aliasing) usually results in a softer image than the

26MP 6K oversampling cameras for 4K video, DR being equal?

Yes, with Bayer sensors, 1:1 readout is always softer than oversampled readout. Strong OLPF that combats moire can make it even softer.

1 minute ago, Jay60p said:

So there is no oversampling for 4K. How could this match the IQ of the X-T3/4 and all the other 24-26 megapixel cameras for 4K video?

My X-T3 is oversampling from 6K to 4K.  Seems to me the 422 or 16 bit raw wouldn't matter much if the lack of oversampling results in a softer

image.  Or is this new 12 megapixel sensor a major new technology?

Oversampling minimises moire/aliasing, it does not directly correlate to better DR.

AP's new logo after this:

I’ve seen one of the cooling solution prototypes Smallrig is designing for R5, it’s not only a cage like in the picture.

They had a 60mm Noctua fan mounted side by side at the bottom of the camera, with a battery plate at the side that can also power the camera.

Smallrig will actually release a cooling accessory for R5, it's not just a cage.

And the lens in the picture is Sigma 24-70/2.8 RF mount.

It's a 1.1x crop from 4240-pixel wide sensor for 1:1 4K UHD readout, this crop is there from 24p all the way up to 120p.

8K RAW frame grab, shot with animal eye AF.

8 hours ago, Andrew Reid said:

BS maybe.

"In-sensor scaling" = pixel binning or oversampling...That is the question. Because binning, skipping, implies you are not sampling all the data, but with averaging and combining data from all photosites, the story is very different.

I am curious how the 12K RAW stuff does not infringe on RED's patents BTW.

Changing the CFA is enough?

I doubt that.

Well put.

BMD patent document on the URSA 12K sensor design:

https://patentimages.storage.googleapis.com/88/08/39/428dbbc9e5dfca/US20190306472A1.pdf

Actual CFA pattern:

In the document, it specifically discusses colour-aware pixel-binning to increase sensitivity and double framerate, at the expense of lower resolutions. So the "oversampling" mentioned in marketing is without doubt BS, at least for full FOV framerates higher than 60fps.

Because if it oversamples from full 12K, then the oversampled modes cannot exceed 60fps which is the full readout max fps.

Maybe there's a feature similar to R5's "4K HQ" mode.

This is the CFA pattern of the URSA 12K sensor:

6x6 block instead of traditional 2x2, 18 RGB pixels and 18 white/transparent pixels, which improves SNR a bit but reduces resolution.

So the optimal shooting mode will be 4K (full RGB info without interpolation from 3x3 pixel-binning), 8K will be softer, 12K 1:1 will be very very soft.

Yes I invented this 12K sensor, which is precisely the same size as RED Komodo at 27.03mm x 14.25mm, and exactly double the horizontal and vertical resolution.

And who made the RED Komodo sensor?? 😉

Ok fine, this 12K sensor is a custom version of Canon's 120MXSC, operating in an ROI mode https://canon-cmos-sensors.com/canon-120mxs-cmos-sensor/