F35 vs. Blackmagic 4K

F

Film Scientist

Active member
I wanted to compare how close the latest entry level 4K compared to the best 1080p from yesteryear (F35),
and whether keeping the F35 was still worth the effort with all these newfangled 4K cameras coming out.

My intention was to do a simple real world test to see how close I could get both cameras to match in 1080, and gauge the major differences in those pixel peeping moments.
Please take the images below with a grain of salt. They are not scientific by any means. I just put the two cameras side by side and pointed at some fruit.

I didn't use the same lenses for my test because the mounts were different, but there was some good glass on both cameras.
In this test the Blackmagic had a 85mm 1.2, and the Sony a 55-105mm Zeiss VP (@95mm) (both matched to approximate field of view.)
The Canon lens was stopped down to 2.2, and the Zeiss fully open at 2.2. Blackmagic was @400 iso, F35 @ 450. Both were set to 5600k. with late afternoon sunlight from the window as the only key.

It's interesting to consider their lens combinations as a system.
The Canon 85mm lens was designed for full frame still sensor, so despite being stopped down 2 stops, to my eye it still felt a little bit softer.
Note, the focus fall-off on the red mesh is slightly different based on the focal length differences. Pay attention to the wood grain of the table, and the tangerine skin for sharpness.
When comparing the downscale of both @ 1080p the F35/Zeiss still felt slightly sharper.
It's a pretty good testament to the fact that the best of yesterday compares very well to the latest from today.

It also shows it's not always about best technology on paper spec sheet, but the right combinations and applications of it.

See the images below, I can happily say the F35 more than holds its own, I actually prefer it.
The F35 seemed to render and capture the colors I observed more accurately.
The combination of the F35/Zeiss yielded that right shade of orange 'out of the box'.
The Blackmagic felt a little duller in terms of color (some of this is was due to 4:2:2 ProRes).
While the Blackmagic could be color corrected to match more closely, on F35 the information was just there, it didn't need to be invented or tweaked.

What this test showed me was the simple fact that both cameras are pretty awesome, both yield awesome 1080p.
The F35s full RGB sensor is really sensitive to subtle color gradations. The Blackmagic yields a highly correctable image that can inter-cut with any camera.

In the future, I am curious to see how the 4K raw on the Blackmagic will compare to the F35s uncompressed RGB stripe sensor.
Its a great to live in a time where we have access to such amazing quality.

BM 4K
4K_graded_00000.jpg

F35
F35_graded_00000.jpg

BM 4K original (film gamut - log)
4K__scaled_untreated_00000.jpg

F35 original (S-gamut - log)
F35_Untreated_00000.jpg
 
Tried my own grade.
Focused on matching wood in shadow and wall for both shots to see how it would effect color on oranges.
For a long time I have felt that CMOS cameras treat skin tone as essentially one color, and have remarked how the F35 allows skin to actually change color/tint from shadow to highlight.
In the real world, skin color is a combination of blood and pigmentation. The blood shows up from sub-surface scattering, where our skin, like wax, allows light to enter and bounce around inside before exiting again.
When skin reflects light at shallow angles (specular highlights), the amount of pigment colored light overcomes the amount of sub-surface scattered light (red), tinting the skin away from red towards yellow.
In shadow, where specular light is at a minimum, the ratio of sub-surface light to pigment light is much higher, resulting in redder (blood) tones.
I think these 2 grades demonstrate this even with the oranges.
Compare the 2 images and focus on the deepest shadows of the orange skins. In one picture, the shadows go more red, and the extreme highlights move toward yellow. In the other, the same color seems to be maintained throughout, almost artificially.
The 4K image treats the color of the orange as being consistent all the way from shadow to highs, ignoring that subsurface scattering would actually effect color the most in shadow. I think this is the result of using a debayering system to achieve color instead of sampling it accurately.
The F35 allows a distinct change in tint, going from reddish shadows to yellow highlights.
At least that's what my eyes are seeing.
Granted, this is really a very subtle observation, but I think it's real.

Love to be proved wrong, if this is just my imagination . . .



Full size here -
http://www.sandboxproduction.com/F35_Clean.png
http://www.sandboxproduction.com/4K_Clean.png
 

Attachments

  • F35_CleanCropped.jpg
    F35_CleanCropped.jpg
    16 KB · Views: 1
  • 4K_CleanCropped.jpg
    4K_CleanCropped.jpg
    17.5 KB · Views: 1
I love testing and I love seeing comparisons! Keep them coming!

I owned the BMPC4k for 2 days before returning it :) Motion and colours on the BMPC4k are really nice. What isn't no nice is the limited highlight room, limited dynamic range in general and the fixed pattern noise that shows up in the shadow areas...

Would love to see a high contrast daylight scene and a dark and murky tungsten scene! :)

As for detail, I think the BM has the edge in terms of zooming in and reframing. But the F35 seems to render detail in a unique way. It seems to accentuate fine detail and give it a very crisp matte dry look.
 
Hi Sandbox,
It's not your imagination, I see the same effects too. The orange appears as one color, minimal color variation on the Blackmagic, like an 'orange blob'.
The F35 nailed the color of the real orange while picking up subtle variations in hue and specularity.
 
Let's see some comparisons of skin tones without labels and see who can tell which is which. Just for fun.
 
Interesting to note also you're comparing one lens stopped down two stops, with one wide open and the wide open one still appears sharper.

Shows just how good proper Zeiss Glass and the F35 are
 
Very interesting findings Film Scientist. But I have to disagree the F35 is the best 1080p from yesteryear. I believe it's the best 1080p camera ever made.

I can't say I'm surprised though. I've been suspecting that about the only thing the BMD 4K had on the F35 was more resolution. But now we see that the detail is not even all that much.

What would be interesting to see would be a 1080p crop from the original 4k frame. If the scaled down version already looks softer than the F35 I would say the 1080 crop would look even worse. Being able to zoom in is the first thing 4k advocates in this 1080p world bring up to justify 4k. If it doesn't hold up there isn't much of a point.


Joncarr said:
Let's see some comparisons of skin tones without labels and see who can tell which is which. Just for fun.
Click to expand...

If it's direct sun light on skin it's pretty easy to tell the smooth and natural F35 highlight roll off from the BMD, the F55 or the REDs.
 
I wonder if the softer look is due to the filtration needed (the nyquest frequency) for a sensor with a color frequency (R & B) that is half of its luminance frequency (G). In effect, it's a 2K camera in R & B, and a 4K camera - with a 2K anti-aliasing filter - in G.
Am I right?
Whereas with the F35, the filtration is matched perfectly for all the colors.

It's surprising to me how differently the F35 handles light in the smallest details.
The 4K looks waxy. With the F35, each bump on the orange skin gets the perfect amount of "chrome" as it were. This also adds to the appearance of sharpness, due to a higher local contrast in each texture.

I've adjusted the image further, to match the values for the table and wall more closely in the crop.

Full-rez here http://www.sandboxproduction.com/OrangesStacked.png
 

Attachments

  • OrangesStacked.jpg
    OrangesStacked.jpg
    11.2 KB · Views: 0
Hi Guys,
Below is a link with a focus chart that is a little more apples to apples, surprisingly moiré is pretty minimal on both cameras, pretty solid.
I didn't downscale 4K file this time, both stills are untreated, so you can made your own corrections.
It's nice to have the resolution of 4K, but the great color also really shines through on the F35. :)

I'll try to do a skin test when I have a little more time.

Cheers,
Manuel

http://ge.tt/8vi5umd1/v/0


F35 Focus_00000.jpgBMP4K Focus_00000.jpg
 
the BMC is resolving your chart - you would need to move back so that the chart cannot be resolved to drive moiree..

edit: in fact it is not the right type of chart really so just moving back may not induce moiree - you have to hit the right point, which is done with the sort of chart that has progressively increasing resolution

BTW - Im really interested in this testing - thank you - Ive seen moire explosions on the BMC :)

S
 
Last edited:
Yes, you're right it is resolving the chart at that distance, it was more of a focus test for me than a full on moiré test.
I need to be able to hook the BM4K to a large monitor with 1:1 to properly focus at a larger distance, so I am trying my best to eyeball it with the little lcd.
I'll try to redo it next week as a torture test, recording something with the chart and a cell phone to expose the moiré.
 
Not to go too off topic, but was watching the 2011 Zacuto shootout, and had to grab frames from F35 and Kodak 5219, and see how hard to match the F35 up to it. No special mattes here. Just some adjustments in saturation and contrast, with VERY little adjustment to color. And that's with heavy compression as part of the pipe-line. I'm surprised so many of the same tones appear in both pictures, especially considering how different the oranges looked in Film Scientist's test, once neutral colors were matched. Here, walls and furniture look pretty close, and skin only differs in exposure latitude.
Full rez here: http://www.sandboxproduction.com/Zacuto5219vsF35.png

Zacuto5219vsF35.jpg
 
Hi Sandbox,

I agree with your assessment. Blackmagic is a Bayer 4:2:2 4K camera. The F35 is a true 1080p 4:4:4 camera.
The F35 dyes on the color filtration seem to be very accurate, as the F35 chip is picking up the subtle hues of the tangerine skin correctly. FYI the skin is muted orange, (as the F35 captures) not a deeper hue.
The 4K cam does feel 'waxy' and somehow captures a deeper orange with too much red, which is not true to life. This is probably a result of the color interpolation from the red and green pixels nearby.

One thing to note when grading is that the back wall is a creamy off white color.

Your Zacuto framegrab definitely shows how close they were able to match Kodak 5219 stock, pretty sweet.

Sandbox said:
I wonder if the softer look is due to the filtration needed (the nyquest frequency) for a sensor with a color frequency (R & B) that is half of its luminance frequency (G). In effect, it's a 2K camera in R & B, and a 4K camera - with a 2K anti-aliasing filter - in G.
Am I right?
Whereas with the F35, the filtration is matched perfectly for all the colors.

It's surprising to me how differently the F35 handles light in the smallest details.
The 4K looks waxy. With the F35, each bump on the orange skin gets the perfect amount of "chrome" as it were. This also adds to the appearance of sharpness, due to a higher local contrast in each texture.

I've adjusted the image further, to match the values for the table and wall more closely in the crop.

Full-rez here http://www.sandboxproduction.com/OrangesStacked.png
Click to expand...
 
I really don't understand why people keep being "surprised" at the F35 being so sharp. After all, ignoring the sensor layout for a second...

F35:

5760 x 2160

BMC 4K:

3840 x 2160

....

I should also point out the F35 can retain its full resolution at 50P which is something the BMC4K and even the original RED could not do.

A friend recently had the bmc4k on set and told me the native ISO is actually around 200 - is that right?
 
You must log in or register to reply here.
Back
Top