New Blackmagic Camera July 16th?

Geoff Boyle posted his under/overexposure test on a production version of the camera with some short comments:

https://vimeo.com/466793043

"This series of exposures goes +/- 6 stops. One rune from T2.8 to T22 was done with filter one and then a second run of the same stops with filter 4.
The 2 series were matched in Resolve in ACEScct with highlight recovery enabled at the "correct" exposure level using CT and Hue only. The other shots in each sequence had this basic grade applied and then the Kodak Grey was set at 488 in every shot using luminance offset only.
I believe that this camera is comfortably usable from -3 to + 4.5 stops with very simple grading and a little noise reduction on the underexposed shots."
 
Teddy_Dem said:
I believe that this camera is comfortably usable from -3 to + 4.5 stops with very simple grading and a little noise reduction on the underexposed shots."
Click to expand...

I haven’t had opportunity to see the video test yet, but if you’re estimate is correct that is excellent latitude.
 
Samuel H said:
If he didn't change anything important in the stest (and I see some changes but nothing really major), he's rating the BM-12k half a stop better than the Alexa Mini.

Here's the Alexa Mini test
https://vimeo.com/280683552
And here's where he said that was good until +5 and -2
https://cinematography.net/CineRant/2018/07/30/personal-comments-on-the-2018-cml-camera-evaluations/
Click to expand...

OK, now that I've watched the clips closely... I think he's being very generous to the BM-12k, treating it as the $10k camera it is, and accepting a lot more noise than he did on the gold standard Alexa Mini.

And still... looking at the clips, I think the BM-UMP12k is perfectly fine at +4, and has acceptable noise at -1.5, showing issues only on the very saturated reds. If you're fine with fixing that in post, this would put it just one stop behind the Alexa, and on par with the Varicam LT and the RED Gemini. Very impressive.
 
Teddy_Dem said:
12K URSA Mini Pro VS 8K Canon R5 VS 6K C500 MKII:

https://www.youtube.com/watch?v=3JGvfbJXZkk
Click to expand...

That is helpful. Cuts through a lot of the marketing and "wow" factors.

Just seems like we can't get into the theoretical aspects of sensor design yet, and bigger pixels and better design still win out, within a bell curve of parameters.

For example, by the time the moire is fixed and denoise is applied on the 12K, why not just have a sharp 8k or 6K camera? Or what advantage is the BM12K at 4K, if the compression ratio has to be 5x's or lower?

That all said, and semantics or technical arguments aside, the BM12k is still producing lovely images. Has an interesting look in the grain. Seems IR pollution is effecting synthetic black material, and moire would be the main fix required. Hope I get to shoot on one sometime. Be able to document something in12K motion is kind of awesome, but perhaps the R5 is good enough for that?
 
James0b57 said:
That is helpful. Cuts through a lot of the marketing and "wow" factors.

Just seems like we can't get into the theoretical aspects of sensor design yet, and bigger pixels and better design still win out, within a bell curve of parameters.

For example, by the time the moire is fixed and denoise is applied on the 12K, why not just have a sharp 8k or 6K camera? Or what advantage is the BM12K at 4K, if the compression ratio has to be 5x's or lower?
Click to expand...

500
http://yedlin.net/ResDemo/

James0b57 said:
Seems IR pollution is effecting synthetic black material
Click to expand...

Did he change shirts at 9:03? Did the dog get dipped in purple ink?
 
James0b57 said:
That is helpful. Cuts through a lot of the marketing and "wow" factors.

Just seems like we can't get into the theoretical aspects of sensor design yet, and bigger pixels and better design still win out, within a bell curve of parameters.

For example, by the time the moire is fixed and denoise is applied on the 12K, why not just have a sharp 8k or 6K camera? Or what advantage is the BM12K at 4K, if the compression ratio has to be 5x's or lower?

That all said, and semantics or technical arguments aside, the BM12k is still producing lovely images. Has an interesting look in the grain. Seems IR pollution is effecting synthetic black material, and moire would be the main fix required. Hope I get to shoot on one sometime. Be able to document something in12K motion is kind of awesome, but perhaps the R5 is good enough for that?
Click to expand...
Well, the whole down sample thing is somewhat murky waters, usually people want a 2K final image to originate from a 3K sensor, and 4K from 5.2-6K to deal with the de-bayering quality loss. However since this one doesn't de-bayer, it probably has another set of "best practice" downsampling. I.e. maybe the 4K ProRes version is sharper and have better pixel quality than the 4K raw, my thinking being that the 4K ProRes might originate from 12K downsample or 8K downsample to get basically perfect RGB-values, but a 4K will not be able to create "perfect" RGB-values in 4K but closer to 2K (since it is just using less pixels from the start, at least if doing faster framerates). Also, seems like their compression ratio might bugged as of now, in the video above they talked about 12 to 1 compression in 4K on the 12K camera was creating sub-par quality, and one had to choose at least 5 to 1 in 4K to get good quality (whereas in 12K and 8K the 12 to 1 ratio created completely fine IQ).

I myself ain't that interested in 12K, but the ability to maybe shoot anamorphic in good 4K or 8K is very aluring.
 
James0b57 said:
That all said, and semantics or technical arguments aside, the BM12k is still producing lovely images. Has an interesting look in the grain. Seems IR pollution is effecting synthetic black material, and moire would be the main fix required. Hope I get to shoot on one sometime. Be able to document something in12K motion is kind of awesome, but perhaps the R5 is good enough for that?
Click to expand...

Even though the Canons had good color (especially the C500 II) I did prefer the colors coming out of the 12K.

We'll have to see about that IR pollution. I'm wondering if the internal NDs were used in that shot with the synthetic shirt. I hope that Blackmagic would have figured it out by now, since every other high end camera seems to have IR pollution solved.

The moire is a little more understandable, since this is such a high megapixel sensor, but it's still troublesome to see. It's annoying to have to rely on a small third party to develop an effective OLPF or to always have to have an IR cut filter in front of the lens.

It was hard to come to a good conclusions about sharpness because the pipelines were different and you had two different people working on the images. I would have liked to see the 12K compared with no denoising and with denoising without sharpening. As it was, I felt that the sharpening they applied to the 12K was too high. It would have also been helpful to see the Canon cameras denoised and sharpened to see how that stacked up. (Ideally they would have shot both cameras at resolution charts to see what settings on each applied an equal amount of overall sharpening.)

It also would have been nice to see just a basic under/over exposure test as well and then that denoised. The advantage of the white photosite is that more total light is being let in, but at an individual photosite level, the smaller photosites on the 12K are noisier than the bigger photosites on the Canons. But, higher resolution images should denoise better than lower resolution images.

I would hypothesize that when temporal denoising is taken into effect (especially when shooting at lower compression levels) you'll see a much cleaner image that would reach farther into the shadows (increasing dynamic range) than a comparable 8K camera. But that remains to be seen.
 
Tom Roper said:
Did he change shirts at 9:03? Did the dog get dipped in purple ink?
Click to expand...

Obviously the synthetic black material would be the shirt at 9:03. But the dog, the backpack to the right and his watchband all looked acceptable to me - not much of a shift at all. The dog looked ever so slightly more purple, but not like the shirt did.
 
To be fair, I see some green/cyan hues to some of the R5 footage. It’s hard to know what’s responsible, the cameras or the lut/grades.
 
Teddy_Dem said:
12K URSA Mini Pro VS 8K Canon R5 VS 6K C500 MKII:

https://www.youtube.com/watch?v=3JGvfbJXZkk
Click to expand...

I don't know what they were thinking shooting the Ursa at 12:1 (where the C500ii shoots 3:1), it also appears they've added digital sharpening as well, which skews things considerably too. Putting so much effort into these tests, and then compromising the results from the get-go is a real shame.

The moire from the Ursa 12k seems just as problematic as ever for Blackmagic (and it'll prevent me considering it for anything, at least until there's a third-party option for it).
 
High1ander said:
Well, the whole down sample thing is somewhat murky waters, usually people want a 2K final image to originate from a 3K sensor, and 4K from 5.2-6K to deal with the de-bayering quality loss. However since this one doesn't de-bayer, it probably has another set of "best practice" downsampling. I.e. maybe the 4K ProRes version is sharper and have better pixel quality than the 4K raw, my thinking being that the 4K ProRes might originate from 12K downsample or 8K downsample to get basically perfect RGB-values, but a 4K will not be able to create "perfect" RGB-values in 4K but closer to 2K (since it is just using less pixels from the start, at least if doing faster framerates). Also, seems like their compression ratio might bugged as of now, in the video above they talked about 12 to 1 compression in 4K on the 12K camera was creating sub-par quality, and one had to choose at least 5 to 1 in 4K to get good quality (whereas in 12K and 8K the 12 to 1 ratio created completely fine IQ).

I myself ain't that interested in 12K, but the ability to maybe shoot anamorphic in good 4K or 8K is very aluring.
Click to expand...
Note that given the sensor readout times of the URSA Mini 12K, it seems like the camera only does a full readout of all the sensor pixels when shooting in 12K: http://www.dvxuser.com/V6/showthrea...issue!&p=1986842959&viewfull=1#post1986842959

When recording in 8K it seems to be doing some kind of pixel binning or other on-sensor scaling approach, as the sensor readout times are much faster than in 12K. The 8K and 4K (full field-of-view) recording modes both have the same sensor readout times, which probably suggests that an 8K sensor readout is done in both cases, with in-camera downscaling to get to 4K.

From the video it seems that it is possible to shoot in 4K RAW on the URSA Mini 12K without a sensor crop. But assuming this involves some downscaling of a higher resolution sensor readout, is this really RAW any longer? Anyway, it's possible that this downscaled RAW (however that is constructed) is part of what led to the RAW compression ratio issues when shooting 4K. It would be interesting to know if the 6K or 4K sensor crop modes share the same problems.
 
You must log in or register to reply here.
Back
Top