BMPCC 4K dynamic range allocation

[Guillaume Cottin]

Hello everyone.

 

Reading the manual of the Blackmagic Pocket Cinema Camera 4K (what a long name…) I stumbled upon this very interesting graphic:

 

 

As we can see, the curve resets at 1250 ISO to adapt to the new sensor amplification. We can also see that the dynamic range is inferior at the 3200 ISO setting and beyond which is not surprising.

 

What I want to talk about here is the significant unbalance between highlight and shadow latitude especially at lower sensitivities. Shooting at 100 ISO for example, feels like a terrible idea with only 2 stops of DR in the highlights, and a ridiculous 11.1 stops under middle grey.

I haven't seen or read anyone talk about it online yet. Maybe people just always ETTR and may therefore never notice this as they never refer use middle grey as a reference.

 

Rating the camera at 1000+ ISO (400 ISO native), gives the more balanced DR at the expense of noise. I have noticed some harsh video-like looking clipping in the otherwise great looking demo footage of the camera. Rebalancing the dynamic range by rating it at a higher ISO might help with those highlights. But then Blackmagic specifically advises not to shoot at 1000 ISO and instead switch to 1250 ISO "for cleaner results".

 

 

When the ISO setting is between 100 and 1,000 the native ISO of 400 is used as a reference point. The ISO range between 1,250 and 25,600 uses the native ISO of 3,200 as a reference. If you are shooting in conditions where you have a choice between ISO 1,000 or 1,250, we suggest closing down one stop on your lens’ iris so that you can select ISO 1,250 as it will engage the higher native ISO and provide much cleaner results.

 

So far this is all on paper because I don't have the camera and the footage Blackmagic provides is not sufficient to completely evaluate this topic. So I am wondering if any of the early birds and owners out there already tried "pushing" the sensor, and if the tradeoff of noise vs highlight DR is acceptable when doing so.

[Andrew Schroeder]

Hi, I've been shooting at ASA800 and am rating it for the same (I don't ETTR). The noise levels are very low, lower than the original pocket cam in a shootout (which is why I chose that ASA). But I'm also shooting in-camera 1080p ProresHQ for an Apples to Apples comparison, so the downscaling and compression will play into that.

ASA3200 is grainier and the color takes a hit, but is still very usuable, not unlike the grain on the orignal cam at ASA800.

I'll have to look at it in 4k and at diff.ratings in the near future, I started with comparing to the original as thats been my fav. camera to work with the last 5 years

[Mark Kenfield]

4 stops of overexposure latitude at native ISOs is pretty limited. And even 5 (if you push it a stop to 800) is pretty underwhelming in this day and age.

I'll be interested to test it and find out how far you can push it for a well-balanced latitude without undue noise.

[Guillaume Cottin]

Adding some info. To have a reference, here is the same diagram for an Alexa. (source: Arri). The Alexa has 1.5 stops more DR than the BMPCC4K and its range is better balanced.

 

I would be interested in seeing an over/under key light test at 1000 ISO and exposed for 2000 EI (either with a light meter or by using a custom viewing LUT). If the noise is OK and the image doesn't fall apart, then on paper it should give a nice balance to the BMPCC4K, at +6.3 stops over and -6.8 under.

[Nelson JJ Flores]

Is there any website or chart (besides the BMPCC) that shows the DR of most cameras out now? Kind of like a comparison?

 

Or are they usually listed on the website of the manufacturers?

[Tyler Purcell]

Adding some info. To have a reference, here is the same diagram for an Alexa. (source: Arri). The Alexa has 1.5 stops more DR than the BMPCC4K and its range is better balanced.

It's really not fair to compare a nearly $80,000 camera to a $1300 camera. Arri doesn't have to think about it's chip set, they use the best period and whatever it costs to make, plus profit, is what they charge the customers. Blackmagic has to pick a price first and make a camera for that market, so they have to skimp out on so many things. I mean if Arri even tried to make a $1300 camera, they'd fail miserably.

 

The dynamic range charts also only reference the imager in a "scientifically perfect" setting, not the captured media in post production. So they don't necessarily mean much besides comparing the quality of a (assumed) low-end BMP4k imager vs high-end Alexa imager. Also... to this day, I think most people consider the Alexa's color science and imager, to be the gold standard.

 

It's kind of like comparing a $99 budget phone to an iPhone X. I guess they both do the same thing sorta?

[Guillaume Cottin]

OK, I was really expecting someone to bring up that sort of thing when I added the Alexa chart. I was tempted to write " I am not trying to compare Alexa vs Blackmagic blablabla", but I thought it was so obvious that it would be an insult to the reader's intelligence, so I wrote instead " Adding some info" and "to have a reference point…" which, I thought, already conveyed the idea that I am not directly comparing the two.

But it wasn't obvious enough so let me make it very clear. The subject is how to manage the limited highlight latitude of the BMPCC. The Alexa is the gold standard in highlight latitude. That's why taking that camera as a reference makes sense, but of course whatever tweaking or workflow we are able to invent with the Blackmagic won't match it anyway. Also, the Alexa handles the ISO changes quite similarly to the BMPCC yet with better allocation yet only 5 or 6 stops of overexposure latitude at some settings. That is why it is interesting to see both charts.

 

As for iPhones: isn't there a Godwin law for cinematography that says "As an online discussion on cameras grows longer, the probability of a comparison involving iPhones approaches 1"?

[Tyler Purcell]

As for iPhones: isn't there a Godwin law for cinematography that says "As an online discussion on cameras grows longer, the probability of a comparison involving iPhones approaches 1"?

hahahah so true! :P

[Jay Young]

If that image is correct, whats the real difference between 100 and 1000? Seems to me you would loose more than you might gain.

[Robin R Probyn]

Isnt it odd to have native ISO at 400.. when according to the chart you have only 4 stops over grey.. but 9 stops under !! more than twice the over latitude ..