Ursa 4K in low light

We just launched a video that we shot with the Ursa in a pretty low light environment in Frankfurt's subway system. We didn't use any additional lights except what was on location.

My lightmeter was suggesting an F1.2-F1.8 at ISO 400 / 60 FPS / 172.8deg. I shot mostly around F2 to have better control on the focus with the ISO locked on 400. Some shots are even at 120 FPS, but recovered really well in post.



I didn't intervene to much on the grade, only added some saturation and a little bit of contrast. I was looking for this moody / murky grade, so the footage looked pretty good and consistent in cam.

Looking forward to see your thoughts!

Looks great. I've never had any issue with my URSA in "low" light, even at high frame rates. If you know what you're doing, you get great results like you've shown here.

Looks great. I've also been doing some tests with my URSA Mini 4K and a new Samyang T1.5 lens at ISO 400. The results so far have been very encouraging.

Sent from my SM-N920V using Tapatalk

Looks great indeed. Did you use noise reduction? I noticed noise in her black dress in 2 shots, but none (as far as my bare eyes can see) in others.

Looks great.

Thanks, Joshua!

I've also been doing some tests with my URSA Mini 4K and a new Samyang T1.5 lens at ISO 400

This was actually shot with the 24mm Samyang VDSLR lens. It's softer when wide open than a Zeiss or Canon, but I feel the softness kind of adds to the look when you're not doing mission critical stuff.

Did you use noise reduction?

Didn't use any NR on this one, Ruben. I suspect the noise you're seeing is actually produced by some encoding artifacts from Vimeo or Premiere. But if you let me know which shot you're referring to I can double check on the RAW footage.

This was shot mostly RAW 3:1, BTW. Except the really slow shots, naturally.

It's the shots at 0:24 (standing) and 0:29 (walking) and 0:45 (in metro). I know it's a bit nitpicking, but I was just wondering why these shots look different noise-wise. Especially the shot at 0:29 looks noisier. The blacks in other shots look cleaner.

Its not a surprise for me that big Ursa 4K shooting better in low-light that ursa mini 4K. Theoretically V2 sensor its a improvement to V1 sensor but practically i see thats not truth. Im thinking that big Ursa 4K capabilty to shooting at 120 fps and ursa mini 4k capability to shooting only 60 fps (full mode) make the difference even in low-light situation not just in slow motion.

It's the shots at 0:24 (standing) and 0:29 (walking) and 0:45 (in metro). I know it's a bit nitpicking, but I was just wondering why these shots look different noise-wise. Especially the shot at 0:29 looks noisier. The blacks in other shots look cleaner.

You're right, there's a bit of noise in the dress in those shots:
- I've underexposed the shot at 0:24 protect some of the highlights (reflections) in the back.
- the shot at 0:29 was severely underexposed due to low light, so I upped the exposure in post without NR
- the dress at 0:45 is indeed underexposed, because the light in the cart was simply not enough

Still, the noise on the black dress is pretty manageable either with NR, or even by just lowering parts of the shadows a tiny little bit. There aren't too many shades of black and it wouldn't affect the overall look too much.

Its not a surprise for me that big Ursa 4K shooting better in low-light that ursa mini 4K. Theoretically V2 sensor its a improvement to V1 sensor but practically i see thats not truth. Im thinking that big Ursa 4K capabilty to shooting at 120 fps and ursa mini 4k capability to shooting only 60 fps (full mode) make the difference even in low-light situation not just in slow motion.

I have the V2 sensor on my Ursa, Michael. I can shoot at 80fps / 150fps, but decided not too since it wasn't necessary for the story and the fluorescents were flickering like crazy in the subway stations.

I haven't seen too many V1 sensors, so I can't say if V2 is better or not. But (I'm starting to look like a genuine fanboy by the amount of times I've mentioned this) I'm happy with the capabilities of this camera package.

My advice is, again, to shoot at ISO 400 and up the exposure in post afterwards. Check out the shot below. I know, it's just a still and a totally non-technical approach, but to my eyes, the skin tones (even reds) are holding up well when underexposed by 2.5 stops.

Left is original, Right is post

ursa underexposure.jpg (581.7 KiB) Viewed 2639 times

I have the big Ursa 4K v1 sensor- since I've been using the evf and a waveform I've found that it can work quite nicely at 800 iso- but you can't really raise the black level above what you've shot- a waveform is a great help with this, as is false colour. It rolls off nicely into clip anyway so it's not a major concern.

I used to rely on the fold out screen but I always ended up under exposing my shots, having to raise and introducing noise.

I have the big Ursa 4K v1 sensor- since I've been using the evf and a waveform I've found that it can work quite nicely at 800 iso- but you can't really raise the black level above what you've shot- a waveform is a great help with this, as is false colour. It rolls off nicely into clip anyway so it's not a major concern.

I used to rely on the fold out screen but I always ended up under exposing my shots, having to raise and introducing noise.

I got used with the big screen on the Ursa, Gavin. It does tend to up the blacks by a tiny amount compared to what I see in Resolve on the waveforms... and the image looks clearer on the camera which can be really misleading.

Now I'm also using the BM Video Assist when doing shoulder work for personal projects and you can tune it a bit better from the contrast / brightness adjustments.

Although I'd really like an adjustable false color on the camera to have an exact representation of the levels... The FS monitors help you nail it every single time in that regard.

It's the shots at 0:24 (standing) and 0:29 (walking) and 0:45 (in metro). I know it's a bit nitpicking, but I was just wondering why these shots look different noise-wise. Especially the shot at 0:29 looks noisier. The blacks in other shots look cleaner.

You're right, there's a bit of noise in the dress in those shots:
- I've underexposed the shot at 0:24 protect some of the highlights (reflections) in the back.
- the shot at 0:29 was severely underexposed due to low light, so I upped the exposure in post without NR
- the dress at 0:45 is indeed underexposed, because the light in the cart was simply not enough

Still, the noise on the black dress is pretty manageable either with NR, or even by just lowering parts of the shadows a tiny little bit. There aren't too many shades of black and it wouldn't affect the overall look too much.

Thanks for the heads up. It's indeed a miracle with 2.5 stops underexposure and still looking this good. The result is what counts