Blackmagic 4K URSA Tests Part 1
Previously, we delivered the Canon EOS C100 MK II tests. Now we are providing the Blackmagic’s URSA 4k so that you can see what paint brushes you might want to add to your creative tool box.
Going Inside the Blackmagic URSA 4K Camera for Cinematic Capture
At first glance, the camera looks like many of its competitors, like the Sony F55 or AJA Cion. It is built to look like an ENG camera and posed to enable the documentary market that is thirsting for 4K content. The tests below quickly show you what this camera can and cannot do. Is it a low light shooter? No, but not everything is about shooting in low light scenarios. There are other cameras that do that in spades. Is it a noisy camera? Yes. Let’s take a look at what the URSA does do well.
Blackmagic Design: URSA 4K Tech Specs Sheet
Why I Do Tests
Tests are everything in finding your camera’s soul. What it does well and where she breaks. You know how specific I am with tests and this is what you pay for. These tests take a team of over ten collaborators who give of their services to educate all of you.
They cost a good amount of money to do them right. It takes time, attention to detail and accurate notes that I can pass onto all of you. This test was so much fun because I brought back most of my amazing crew from the Illumination Experience tour to do it. My mission is to turn these guys into great additions to the movie business labor pool. I drive them and push them to be excellent.
OK, here we GO!!!!
Test No. 1 – Looking at IR Pollution
The first test that I always do with a new camera is to verify how good its Optical Low Pass Filter (OLPF) is. The BMCC and the BMPC exhibit strong IR pollution after two stops of ND and the URSA reacted the same way. If you are shooting with this camera during the day time and you use Neutral Density so you are not shooting all your imagery at an F11, then you have to follow this test closely because you will have conditions with normal Neutral Density filters that will bake in an IR polluted look that you will not be able to remove.
I feel you are safe up to two stops of ND, so a WW ND 0.6. After this, you really need to start to use IR NDs to combat all this IR pollution. It affects the skin and green trees first and then drifts into turning your sky purple. We thought it was an awesome look for 1970s found footage. Looked perfect. Putting that into my cinematic memory bank.
Bringing It Back with IR Filtration
We wanted to show you how using an IR filter can help your color grading and assist in not baking the IR polluted look into your footage. We used WW IR NDs from Tiffen for this test and I feel they reacted very well with this sensor.
I felt the Tiffen 2.1 IR ND did not have enough IR embedded into its glass. The 1.5 and 1.8 did a great job counteracting the IR pollution, but the 2.1 did not cut it. Her skin is a weird tone and the background feels like it has this sandy baked in feel.
Test No. 2 – Let’s break the camera!!
This is the most time consuming test. Not many people do the latitude test, but it is the most informative. Finding the sweet spot of your sensor is not done with latitude charts or that the latitude light machine that you see so many tests use. Our work is for you to use in the field, on location or in a studio with extreme conditions.
We are going to start with over-exposing the sensor to push it to its breaking point. What I noticed with this particular 4K sensor was that it doesn’t exhibit the latitude of the BMCC or the BMPCC. I found a useable latitude of about eleven stops. It blows fast, but it has a beautiful roll off into the highlights. I think this is one of the strong points of the camera, so even though it is limited by its latitude, the way it over-exposes feels more like the Arri Alexa. The Canon C300 or C500 clip very fast. It might have 12 stops, but if it looks like video clipping highlights, then what is the point?
I set the exposure on this camera at 18% grey on the False Color chart at 40-45 IRE. Any higher than this, then the camera over exposed very fast. Her skin tones were at 50-60 IRE. Let’s start at the base exposure. My light meter, the Sekonic L-478 DR, reacted perfectly to this camera at 400 ISO.
Up to +1, I feel I could use this intense feeling of light on her face and bring the grade down to hold all detail on her face, still very pleasing. Let’s look at her skin tones especially with this test. This is where the Blackmagic delivers beautiful 12 BIT skin tones that have wonderful depth and dimension. Love it!!!!!
I could live with +11/2 stops over on Eli, but this is where the C500 started to clip unnaturally on the face of the male model on the Need For Speed tests.
Now let’s examine how well this camera rolls off its highlights. With +2 over exposed on the 18% grey card, her skin is right at 100 IRE on the False Color scale, but what I find beautiful about this camera is how her skin doesn’t exhibit that clippy video feel. It is very smooth. I look at this camera and say I love the skin and I could shoot this easily in white cyc situations. You could expose your wall nicely at +2 over and then your subject around +1/2 and rock it out.
At +2 ½ stops, she starts to clip, which looks a little extreme but better than the C300 or C500.
Look how the BMCC cares and exposes beautifully even on +2 ½ stops.You can bring it back and it looks awesome.
Test No. 3 – Let’s Dive into Under-Exposure
What I learned from this test is that it doesn’t exhibit the same underexposing qualities of the BMCC or BMPCC. It falls apart faster. But what I did discover that you could do with this camera is use its ability to handle some underexposure to combat the noise level that you get in low light situations when you slide the ISO up to 800. You will soon see this in our day exterior noise and night exterior noise tests. This camera is very noisy at 800 ISO; it is borderline at 400 ISO. So what I propose to do is to shoot night exterior or interior, low light work at 400 ISO, and then use the ability of Davinci Resolve 11 and the sensor’s under-exposing power to bring the image back up to what looks good without the added noise of taxing the sensor at 800 ISO.
The camera handled -1 stops under exposed very well and retained all detail.
Even -1 ½ stops would work but at -2 stops, she really started to lose BIT depth and became very smoky. Bringing the image back just creates this very weird heightened contrast, which looks forced.
The C500 did not handle underexposing very well either in our Need for Speed tests. After -1 ½ stops it fell apart as well, but you could bring back the BMCC pretty well, again exhibiting the increased latitude of that sensor which boasts 13 stops. I was recently shooting with the Alexa and I love how that camera rolls off highlights and how it handles underexposure, which has 14 stops of latitude. These extra stops are huge for making your image shine. You might not have the resources or a crew to constantly be altering the extreme conditions with nets, neutral density on windows, flying large diffusion frames to take values down. But we are looking at price points, right? The Alexa is $70K and the BMCC is $2K and the URSA is $5K. You have to be realistic with your expectations of a camera at its price point. Period.
In this test with the URSA we took our under-exposure all the way down to -5 stops, but unfortunately for us the SanDisk C-Fast Cards we used did not hold up. One of them corrupted on us and we lost our footage past -2 ½ stops. Due to the time we were running into, we could not go back and reshoot that part of the latitude test.
Stay tuned for Part 2, where we will take the URSA through its paces and show you the camera’s noise levels during the day and night — fill ratio, slow motion and rolling shutter.
Hello Mr. Hurlbut,
Thank you for the awesome post. I love you post about testing sensors and hardware to fit your needs as an artist. I have been doing that a lot more in my work with still cameras to start as I do more still work.
I have been looking into low cost, high value equipment for our needs and the BM line was a consideration. I was told by a friend that his experience with BM files corrupting was an serious issue and with your experience recording with it reminds me of his comments. Do you find the BM hardware to be a serious issue when corrupted recordings are discovered? Should this eliminate a BM Cinema from my consideration?
Hi David. From what we have discovered, I do not believe that it was the hardware of the camera that caused this problem. We have discussed this with Blackmagic and they are looking further into it to verify exactly what caused our card to corrupt. From what my DIT Derek was showing me, it looks like the card failed us, as we had other San Disk cards that day and no other card caused a problem during the day. We even ran the card full and let it continue to record from the 1st slot to the 2nd slot, thinking this could have caused the error, but it did not seem to be the case.
Shane, regarding the Tiffen IRND 2.1 on the URSA, there’s a lengthy thread on BMCuser regarding alternative IRNDs that handle the Blackmagic cameras particular IR wavelength range much better than the Tiffens, for example Hoya ProNDs, Formatt Firecrest IRNDS, and Skier IRNDs. I’ve used the Hoyas on two features now and I’m completely sold. Hope this helps!
Thanks Brandon. I will have to check those out and do a test on them as well. Thanks for the tip!
How cheap Shane?
Just under $6,000 at B&H – http://www.bhphotovideo.com/bnh/controller/home?O=&sku=1044785&gclid=CI6G2q_G1cICFciDfgodjAQAjg&Q=&is=REG&A=details
I use the Tiffen IRND 1.2 and 2.1 circular filters on my BMPC 4k. I feel like they works very well. Without IR and just ND the image goes to complete crap.
It’s the nature of the digital sensor these days, the BMD URSA reacts like it’s other members of the BMD cinema cameras when it comes to IR pollution, but the IR ND cuts it right out.
Another amazing test. Long time reader but first time poster. Couldn’t help myself though after I saw my buddy Cornell Mitchell in the picture at the top of the page. Keep him working! Great talent and amazing heart! Really great of you to push not only these amazing tests but also share the talent behind them.
Kudos Mr. Hurlbut
Thanks for posting Chuck and yes, Cornell came on my Illumination Experience Tour and I loved working with him. He’s a great person and it was a great time bringing back some of the crew from the tour on this test.
Thanks for the great post. I shoot with the BMPC 4k, which I believe has the same sensor as the URSA, so I found your results very interesting; however, there’s one important part that is confusing me, and I would greatly appreciate clarification. In your post you said:
“I set the exposure on this camera at 18% grey on the False Color chart at 40-45 IRE. Any higher than this, then the camera over exposed very fast. Her skin tones were at 50-60 IRE. Let’s start at the base exposure. My light meter, the Sekonic L-478 DR, reacted perfectly to this camera at 400 ISO.”
If I don’t have access to a false color monitor and want to use the Sekonic L-478 for exposure settings, where does this value fall in reference to the Sekonic’s normal incident metering? Did the Sekonic base exposure set 18% grey at 40-45 IRE, or did you take the Sekonic setting and adjust from there to achieve 18% grey at 40-45 IRE value? In other words, If I were to take a L-478 reading off the key would that be the closest to your base exposure, or would I need to compensate +1 to get to your base exposure?
awesome! I bought the tiffen irnd 2.1 because you talked about how awesome they are, I’ve found similar findings though, a little too much color shift. It’s too late to return it, would adding a tiffen t1 potentially work? The benefit of that is I could use it with my other non irnd filters, and potentially get a variable nd filter too. Do you think using a T1 and 2.1 irnd would get rid of the color shift?
I’d love to hear your recommendation, thanks so much for an amazing blog shane!
Hi Shane- I met you a few years ago at Neil Smith’s place in Hollywood. Thanks for all the great info!
In your IR tests with Eli, the sky, building, parking lot stripes, and car highlights have a magenta tint. I did a quick test to try to fix those colors and not change skin tones too much:
Not clear if the jacket was black or navy. With Canon (5D3) and Panasonic cameras (GH4), I don’t have much issues with skin tones and background colors, however with Sony cameras (FS700 and A7S), it can be a challenge (beyond a simple WB adjustment in post).
I understand just about anything can be fixed with secondaries and/or masking, however it would be helpful to learn any tricks which quickly get color pleasing for skin tones and as correct as possible for all scene colors. The goal is to get the shot accurate before coloring for emotion.
Hi Shane, thanks for taking the time to do such extensive testing. I have an Ursa and have noticed a number of quirks and wondered if you’d seen anything similar. At 800 iso I experience a strange horizontal ghosting of highlights across blacks, it appears like a band extending from the lighter source across shadows. At 800 I shot someone in a dark top against bathroom tiles, when I looked back, the white grouting in between the tiles weirdly extends over the foreground subject on the dark jumper. In low light (with everything in the bottom quarter of the histogram) you can see the same thing at 400 (esp if you try to lift anything later). I’ve sent the camera back to BMD and they said it was the same with their test models but would hope for a firmware fix as the BMPC4k doesn’t exhibit the ghosting. At 800 iso looking at the histogram and zebras, it looks like I get more dr in daylight than at 200 or 400, even tho 400 is native. Underexposing by about 1 1/2 stops at 800 in low light produces a fpn. Have you seen this?
Hey Shane, thanks for all your info!!
Hey man I’m just wondering what your position is on T1 filters + straight ND’s vs IRND’s?
I’m curious to know if the CFAST card itself corrupted, or if the camera corrupted the clips itself. I use SanDisk CFAST 2.0 in the Amira and haven’t had any issues (knock on wood). Gone through 100+ rolls on 12 cards.