Blackmagic Pocket Cinema Camera 4K

[seanzzxx]

The one thing that I am worried about when looking at footage from the GH5 and GH5s is edge sharpness. I know this is dependent on the way the manufacturer treats the footage but the Fairchild sensors all had such an amazing 'creamy' look that made me immediately fall in love with the Ursa's when I first rented them. I do not see that on the GH5(s) footage at all, it looks like a digital camera right away. I hope the Blackmagic 'look' was not partially reliant on the Fairchild sensors. It will be a very interesting time when it comes out!

[jonpais]

The GH5 series is constantly being accused of suffering from excessive sharpening, but in fact, they they have considerably less sharpening applied than either the a7 III or the X-H1! Compare the X-H1and a7 III, with a maximum of 17.0% to the GH5s, with a maximum of just 6.2%! And those results are at factory default settings. What people may be seeing is a triple threat of compression, noise and sharpening.

Fuji X-H1:

Sharpening in video is moderate and similar at both low and high ISO: 10.3% and 10.1% overshoot, respectively, along high-contrast edges, and 13.2% and 12.0% undershoot.

Along low-contrast edges, video stills show 12.2% (low ISO) and 12.1% (high ISO) overshoot, and 17.0% and 13.8% undershoot, respectively.

Sony a7 III:

Sharpening in video is moderate and similar at both low and high ISO: 10.3% and 10.1% overshoot, respectively, along high-contrast edges, and 13.2% and 12.0% undershoot.

Along low-contrast edges, video stills show 12.2% (low ISO) and 12.1% (high ISO) overshoot, and 17.0% and 13.8% undershoot, respectively.

Lumix GH5s:

Sharpening in video is mild: 4.4 percent overshoot and 2.6 percent undershoot at low ISO along high-contrast edges, and 1.2 percent and 0.7 percent, respectively, at high ISO.

Along low-contrast edges, sharpening was 6.2 percent (overshoot) and 3.7 percent undershoot at low ISO, and 1.0 percent overshoot combined with 1.3 percent undershoot at high ISO. 

 

 

[deezid]

1 hour ago, jonpais said:

The GH5 series is constantly being accused of suffering from excessive sharpening, but in fact, they they have considerably less sharpening applied than either the a7 III or the X-H1! Compare the X-H1and a7 III, with a maximum of 17.0% to the GH5s, with a maximum of just 6.2%! And those results are at factory default settings. What people may be seeing is a triple threat of compression, noise and sharpening.

Fuji X-H1:

Sharpening in video is moderate and similar at both low and high ISO: 10.3% and 10.1% overshoot, respectively, along high-contrast edges, and 13.2% and 12.0% undershoot.

Along low-contrast edges, video stills show 12.2% (low ISO) and 12.1% (high ISO) overshoot, and 17.0% and 13.8% undershoot, respectively.

Sony a7 III:

Sharpening in video is moderate and similar at both low and high ISO: 10.3% and 10.1% overshoot, respectively, along high-contrast edges, and 13.2% and 12.0% undershoot.

Along low-contrast edges, video stills show 12.2% (low ISO) and 12.1% (high ISO) overshoot, and 17.0% and 13.8% undershoot, respectively.

Lumix GH5s:

Sharpening in video is mild: 4.4 percent overshoot and 2.6 percent undershoot at low ISO along high-contrast edges, and 1.2 percent and 0.7 percent, respectively, at high ISO.

Along low-contrast edges, sharpening was 6.2 percent (overshoot) and 3.7 percent undershoot at low ISO, and 1.0 percent overshoot combined with 1.3 percent undershoot at high ISO. 

 

 

The GH5 is accused for sure. Since the oversharpening in 4K or even HD is just way too much. In 6K there is almost no sharpening at all.
The sharpening happens while downscaling. The more it downscales the worse it gets (6K vs. 4K vs. HD vs HD slowmotion).

Since the GH5s doesn't downscale at all, the problem is basically fixed on this camera.

[seanzzxx]

This comment wasn't aimed solely at the GH5, I talked about those cameras because the S has the same sensor. I own an A7rii and I dislike this same quality about the footage. Surely you would agree that edges look much softer on Blackmagic cameras?

[John Brawley]

Edge sharpness. 

First we should look at the imaging chain. Is there an OLPF ? The absence of one will mean a sharper edge but the increased chance of false colour moire. 

Sharpness settings in camera menus aren’t compare-able I would have thought camera to camera. 

Seems like if edge sharpness is what’s being examined then you’re really talking about “detail” and “coring”. 

These are both setting that are in video encoded products but are not in BMDs camera video processing (except for the broadcast cameras) 

These are post applied video processing of signals that tries to improve edge sharpness.  Some cameras allow the user to adjust these settings.  Some bake them in without any user adjustment.

JB

 

[seanzzxx]

Yeah I'm pretty sure I lack the technical vocabulary to describe what I'm talking about. If these are encoding dependent then that would be very encouraging, since that would mean that blackmagic look will remain intact. I guess that's supported by Steve Yedlins work stating that a sensor-by-sensor look is not determined by the sensor but by the manufacturer's preparation of the sensor data provided.

[webrunner5]

I would think most of the Voodoo Magic sauce is in the processing engine. I mean all but Sony use different manufacturers sensors and they still can maintain the Same look to their products. Sure the look may get refined, but not changed at it's basic core output.

I suspect this new PK4 will have somehow the same look available as the BMPCC had, and a Ursa look thrown in for good measure.  ?

[kye]

Wouldn't any digital sharpening, noise reduction and compression effects be N/A when recording in RAW?

I understand that the OLPF and other optical aspects would of course still apply.

[webrunner5]

Most are some sort of a Tiff file from years ago.

https://en.wikipedia.org/wiki/Raw_image_format

[jonpais]

Looks like sharpening in post can be pretty complicated.

[seanzzxx]

33 minutes ago, kye said:

Wouldn't any digital sharpening, noise reduction and compression effects be N/A when recording in RAW?

I understand that the OLPF and other optical aspects would of course still apply.

From what I understand technically raw can be prepared in any way you like, however the vast vast vast majority of people will load the raw into a standard NLE which will use the manufacturer's recommendations on how the image will be displayed. Someone correct me if I'm wrong.

[kye]

1 minute ago, seanzzxx said:

From what I understand technically raw can be prepared in any way you like, however the vast vast vast majority of people will load the raw into a standard NLE which will use the manufacturer's recommendations on how the image will be displayed. Someone correct me if I'm wrong.

My understanding was that the RAW settings that could be taken from the camera was around things like ISO, WB, etc.  I'd be stunned if something like Resolve would do anything other than the minimums to construct an image from the RAW data, and I can't see why sharpening would be required for that.

In a sense, shooting in RAW is desirable precisely because it removes all the decisions that other people make for you in the other modes, so that you can make them yourself.

[deezid]

2 hours ago, John Brawley said:

These are both setting that are in video encoded products but are not in BMDs camera video processing (except for the broadcast cameras) 

 

 

And that's what makes Blackmagic, Arri and RED stand out from the rest. 

[Anaconda_]

I really hope the answer refers to shooting 120fps.

I was under the impression normal 1080p (up to 60p) was no sensor crop, so 1.9x 'relative to Full Frame 35mm'

[webrunner5]

Well it is a m4/3 sensor. It is going to have a 2X crop or more in Any mode wouldn't it retaliative to FF? But 4X?? Yikes.

[Anaconda_]

5 minutes ago, webrunner5 said:

Well it is a m4/3 sensor. It is going to have a 2X crop or more in Any mode wouldn't it retaliative to FF?

Yes, but that answer, to me, implies that HD has a 2x crop of the MFT sensor, and 4x crop compared to a full frame sensor.

[webrunner5]

5 minutes ago, Anaconda_ said:

Yes, but that answer, to me, implies that HD has a 2x crop of the MFT sensor, and 4x crop compared to a full frame sensor.

God I Hope that is for 120fps, and even if it is, what kind of lens would you Need to shoot that in LoL. Hell even a 7-14mm would not be wide! Gulp.

[Anaconda_]

5 minutes ago, webrunner5 said:

God I Hope that is for 120fps, and even if it is, what kind of lens would you Need to shoot that in LoL.

I've not shot much slow motion, but when I have the cameras have cropped in on the sensor to achieve 120fps. When I did it, I had a speed booster with Canon's 10-18mm, so that did the job.

I've also used the Samyang 8mm fisheye with a dumb adaptor. As you can see in the image below, the crop bypasses most if not all the distortion.

[webrunner5]

1 minute ago, Anaconda_ said:

I've not shot much slow motion, but when I have the cameras have cropped in on the sensor to achieve 120fps. When I did it, I had a speed booster with Canon's 10-18mm, so that did the job.

I've also used the Samyang 8mm fisheye, and with the crop, most if not all the distortion wasn't in the frame anymore.

Yeah but I wonder if this PK4 is going to have all that distortion, vignetting stuff under control electronically like the big boys do?

[Anaconda_]

3 minutes ago, webrunner5 said:

Yeah but I wonder if this PK4 is going to have all that distortion, vignetting stuff under control electronically like the big boys do?

Well, the distortion and vignetting is on the edge of the frame, but if you cropping the edge of the frame out anyway, and using only the middle of the sense, it doesn't matter if you have vignetting. the 8mm fisheye was made for APS-C cameras, so there won't be any vignetting on an MFT camera anyway  I would test when the camera arrives, but I sold the lens a while ago.