Welcome to our community

Be a part of something great, join today!

Register Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Scarlet Dragon OLPF night test in streetlight

I

Ignacio Aguilar

Well-known member
Joined
Oct 30, 2011
Messages
868
Reaction score
6
Points
18
Location
Madrid, Spain
We have just released this test that has been done with our new Scarlet Dragon camera.


https://vimeo.com/138243557

The idea was to do some shots with wide-open lenses in such a bad scenario as a street lit only with streetlights, use color correction -applying the same settings to ALL shots- to make them look as good as possible, and then push the ISO to see how noise and color behave with each OLPF.

A 24mm Ultra Prime lens was used for spherical shots and a 50mm UltraScope (a very rare set of lenses from the late 90's, not the 1950's UltraScopes) was used for anamorphic.

Everything was shot at 5K, 24 fps and 8:1 compression, the minimum allowed by the Scarlet.

ISO was set to 500 (one stop under the measured light level) and then pushed in Redcine-X up to 4000 ISO (800 was the ISO for anamorphic and has been pushed to 3200).
So the "correct" exposure should be ISO 1000 for spherical shots and 1600 for anamorphic.

R3Ds and 2K TIFF files are also available for download: dropbox.com/sh/guf1ku602g0udv5/AADyQ51g2PVWBVKn3LYTFaJna?dl=0

Hope you'll find this test useful!!
 
Last edited:
Ignacio, this is a wonderful test. Thanks so much for sharing. En hora buena!

What I love most about this test is two things.

1) Its a wonderful reminder of the extra hightlight protection in skintone. You can def see that at the higher ISO's on the 24mm test. The lowlight OLPF clips faster (on the face in praticular), whereas the Sktintone OLPF holds these highlights in an admirable, impressive way. With DEB, the differences in blacks are not as much one would think....

2) It's also fascinating to see what the extra 2/3 stop of light btw the two lenses does to the noise. At T2, most of your footage looks relatively noise free, even at high ISO's (3200, 4000, etc). Once you threw on the Ultrascope at T2.5, you can really see a bit of noise creeping in at higher ISOs. At first glance, that 2/3 stop may not seem like a lot...but clearly...it is.

Makes me even think of Weapon. Its supposed to be 1/2 stop cleaner in blacks (my guess is it's a bit better than this). With your test, we are seeing immediately how significant even a 1/2 stop of light is...especially in lowlight.

Thanks again for the test. Really appreciate it.

best
 
Great test, thanks for sharing!

Nick Morrison said:
1) Its a wonderful reminder of the extra hightlight protection in skintone. You can def see that at the higher ISO's on the 24mm test. The lowlight OLPF clips faster (on the face in praticular), whereas the Sktintone OLPF holds these highlights in an admirable, impressive way. With DEB, the differences in blacks are not as much one would think....
Click to expand...

I think you may be attributing this to the wrong thing. They just raised ISO, which is a boost through metadata. If the camera wasn't clipping at 500 ISO, and the lens was at 1.9, then the camera certainly wouldn't be clipping at 4000 ISO, as the f-stop was still at 1.9 and the amount of light hasn't changed in any way.

If anything, it's a difference in color science.
 
Niko Pueringer said:
Great test, thanks for sharing!



I think you may be attributing this to the wrong thing. They just raised ISO, which is a boost through metadata. If the camera wasn't clipping at 500 ISO, and the lens was at 1.9, then the camera certainly wouldn't be clipping at 4000 ISO, as the f-stop was still at 1.9 and the amount of light hasn't changed in any way.

If anything, it's a difference in color science.
Click to expand...

I didnt mean "actually" clipping. Its obv not really clipping. But when viewed at 4000 iso, then the resultant image on LLO was clipping in the highlights...whereas the Skintone was not.

And Im not sure it is just color sciene. Both olpfs are clearly different, and map the dynamic range uniquely - the lowlight appears to add DR to shadows, whereas Skintone adds DR to highlights. Certainly how I understand it. And seems in keeping with most exposure strategies (800 for LLO, and 320 for Skintone).
 
I suppose the noise from the STH OLPF at ISO 1000 in the 24mm shot is similar to what you would normally experience at ISO 2000 had the shot not been underexposed by 1 stop. The noise at ISO 1000 just doesn't bother me, especially for the shooting circumstances and considering the fact that there's a 1 stop deficit. I still fail to see what all the fuss was about.

I'm curious why you didn't change the shutter to 1/24 (or 360°) in order make up that missing stop since the scene was basically static?
 
Brian Boyer said:
I'm curious why you didn't change the shutter to 1/24 (or 360°) in order make up that missing stop since the scene was basically static?
Click to expand...

Yes, you're right and I could have done that for this test, or even shoot 12fps. But I was trying to test how the camera would perform if I had to shoot a real scene in that lighting situation (or other with very low light levels and no chance to raise them) and I just don't like the look of a 360 degree shutter. So in that situation, if I couldn't get a faster lens, I'd prefer to increase the ISO even though it would show some more noise. I find myself the 1000 ISO low-light shots pretty acceptable considering there's just 5 footcandles of streetlight.
 
Last edited:
Nick Morrison said:
2) It's also fascinating to see what the extra 2/3 stop of light btw the two lenses does to the noise. At T2, most of your footage looks relatively noise free, even at high ISO's (3200, 4000, etc). Once you threw on the Ultrascope at T2.5, you can really see a bit of noise creeping in at higher ISOs. At first glance, that 2/3 stop may not seem like a lot...but clearly...it is.
Click to expand...

Also bear in mind that 5K 6:5 is a format that uses less pixels than 5K full. What astonished me was the fact that the Ultra Prime lens sees much more into the blacks; there's a lot of detail hidden there in the R3D's that the anamorphic lens can't even register, even though the exposure had been compensated. I guess modern coatings are the reason for that.
 
Thanks for the test. I think you shouldn't use DEB with the LLO as it introduces some green tint to the picture (and skin). Only use it with STO.

Pat
 
Patrick Tresch said:
Thanks for the test. I think you shouldn't use DEB with the LLO as it introduces some green tint to the picture (and skin).
Click to expand...

Yes, that's clearly noticiable because we just used the same color correction for both OLPFs, but I think this green tint can be easily avoided in post. A certain amount of green was removed anyway (-15 tint in Redcine-X) because the LED streetlights had that color.

Only use it with STO.
Click to expand...

What's RED's word on that? I believe D.E.B. really helps also with the low-light OLPF for very high ISO ratings.
 
Ignacio Aguilar said:
Yes, that's clearly noticiable because we just used the same color correction for both OLPFs, but I think this green tint can be easily avoided in post. A certain amount of green was removed anyway (-15 tint in Redcine-X) because the LED streetlights had that color.



What's RED's word on that? I believe D.E.B. really helps also with the low-light OLPF for very high ISO ratings.
Click to expand...

From my experience the DEB is great on the STH and it's better to use Denoise slide with LLO (if you want to use RCX as your denoiser).
The DEB corrects RED channel noise in parts of the picture introduced by the STH.

This test would be interesting to see with Neat video used on both with no NR applied. The truth is, you grade what is in front of you and you make it look its best. The LLO does give you gains in the lower stops at a cost of protecting the highlights, the STH gives you really nice skin and highlights at the cost of that lower stop. All depends on your setup.

This is an "extreme" test, but once any kind of lighting is introduced then both with perform well depending on the approach taken by the DP.
 
DEB was introduced with the second Olpf Red did for the Dragon which is now called STO (version 1), which provided some red spots in the shadows. Those spots aren't present with the LLO and therefore isn't necessary to use DEB. DEB even lower color definition and adds some green tint to the picture.

Don't use DEB with LLO and do as David said : use a small amount of neat denoiser if you push more than 3200 ISO.

To test Olpf and noise/clipping I would have exposed correctly every shot changing shutter speed and pushed it in post by adjusting ISO accordingly.

200 ISO, 1/25 to 6400 ISO, 1/800

Pat
 
I just wasn't interested in any other shutter speed than 172.8 or 180 degrees, as I Don't enjoy using long or short shutters -unless I really have to, or the director is after that effect- in a real shoot. My goal was just to torture the camera with each OLPF and see how it behaves with ISO pushings, D.E.B. so if I ever have to shoot in similar conditions, I already have a clue to approach them.
 
Ignacio Aguilar said:
I just wasn't interested in any other shutter speed than 172.8 or 180 degrees, as I Don't enjoy using long or short shutters -unless I really have to, or the director is after that effect- in a real shoot. My goal was just to torture the camera with each OLPF and see how it behaves with ISO pushings, D.E.B. so if I ever have to shoot in similar conditions, I already have a clue to approach them.
Click to expand...

I can def see the value in "real world" tests!

Best
 
Ignacio Aguilar said:
I just wasn't interested in any other shutter speed than 172.8 or 180 degrees...
Click to expand...

You'r probably right not to change shutter speed to lower exposition as it also changes the sensor specs and reaction to noise (as we usually don't perform a balck shade for every shutter speed in those kind of tests).

Pat
 
Ignacio Aguilar said:
My goal was just to torture the camera with each OLPF and see how it behaves with ISO pushings, D.E.B. so if I ever have to shoot in similar conditions, I already have a clue to approach them.
Click to expand...

YES. This! Magnifico Ignacio! Thank you for doing that work for us. This was very informative, and I was quite impressed with how well the camera held up.
 
You must log in or register to reply here.
Back
Top