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 Upgrade - Noise issues

G

Guilherme Belchior Afonso

Well-known member
Joined
Feb 4, 2013
Messages
70
Reaction score
0
Points
0
Location
Lisbon, Portugal
Website
vimeo.com
Hello everyone,

I recently upgraded my Scarlet to the Dragon sensor and I´ve made some tests with the STH OLPF and the LLO OLPF and I´m not that impressed. The colours look good and the extra fps are very welcome. But the noise doesn´t look that good. I have already read about issues with noise in other threads but I´m still not clear about what is causing it. I´m getting a lot of color noise in the dark areas. The LLO OLPF is a bit better but still...
I use 1.8 redmags and I know that the minimags are faster, but can I get less compression redcode levels with them in Scarlet D?
Looks like the Redcode options in Scarlet D are very high for the sensor to have a clean image.

I have some contax Zeiss lenses and when shooting at 6K 12 fps they look less sharp than they did at 5K in the MX. Is there any limitation with these lenses at this resolution?
I use the leitax adapter for the lenses with Canon mount, do you guys recommend any other adapter? Or should I change my lenses set?

I made the update just to keep it up with RED´s technology and I don´t regret but I feel like it was to much money for the upgrades and I guess that only in the Epic D with lesser redcode compression the changes are truly noticeable from the MX.

Anyone have some advices?

Thanks
 
I think your sharpness issues at 6k are mostly compression related. We noticed in testing that 4k with lower compression looked nearly as sharp as 6k at lowest compression. And 5k was better than both at lowest compression
 
Guilherme Belchior Afonso said:
In this computer I just have this one. It´s not the best example but have a look Luigi.

Link for download --> http://we.tl/L1mQ3VhcXv
Click to expand...

Maybe I understand what you mean.
Are shooting at 16: 1 compression ratio, this can make things noisy. (especially in dark areas)
Try to make a test with normal speed and compression rate the lowest possible, you will get more "clean images". I usually shoot at 5: 1 to 8: 1, do not go over, just in case of slow.
 
Guilherme Belchior Afonso said:
...I use 1.8 redmags and I know that the minimags are faster, but can I get less compression redcode levels with them in Scarlet D?...
Click to expand...

There's no difference in available Redcode levels when using Scarlet D with either the 1.8" Redmags or Mini-Mags. The lowest setting for both (at 24fps 5K Full Screen) is RC8:1.

Edit - Or as Mike says, RC7:1 @5K Widescreen. Either way, should be noticeably better than RC16:1, which is the lowest compression available when using the max frame-rates (48fps@5K and 60fps@5KWS).
 
Last edited:
I took a look
1) The R3D is 5K 16:1. You can't expect it to be very clean with compression that heavy with so much detail in the scene (trees, bushes, etc.)
2) Although the highlights are clipped in the middle, the trees are not exposed with much light.
3) STH has been said to be noisy past ISO 800.
4) Is your lens known to have chromatic aberration at whatever aperture you used? I see purple fringing in the high contrast area of the clipped highlights/tree edges.
5) In RCX, use the "Pick WB" and choose the clipped highlights. That will make the colors more accurate.
 
Hey guys,

Thanks for the advices! This wasn´t the best example as it was 16:1. Here I only have a few shots, this one is 6K 6:1 and it still has some noise that I didn´t expect and is not that sharp as I wanted. --> http://we.tl/7JzhnNRxce

I´ll make some more tests with better settings to see the differences.

Bob Gundu said:
What F-Stop was your lens?
Click to expand...
-I don´t recall the F-Stop but I think around 5,6.
 
Guilherme Belchior Afonso said:
Hey guys,

Thanks for the advices! This wasn´t the best example as it was 16:1. Here I only have a few shots, this one is 6K 6:1 and it still has some noise that I didn´t expect and is not that sharp as I wanted. --> http://we.tl/7JzhnNRxce

I´ll make some more tests with better settings to see the differences.

-I don´t recall the F-Stop but I think around 5,6.
Click to expand...


I'm not really seeing a sharpness issue and I'm not seeing any crazy noise really at ISO 1000 here.

phfx_forGuilherme.jpg


phfx_forGuilhermeCC.jpg
 
I made a black shading before this shooting.
I guess what maybe is making some differences in my viewing is that since I´ve made the upgrade I also bought a 5K imac and with more pixels at the same size I see more noise. And that is maybe confusing me. Is that possible?
 
The sharpness is there and the noise levels are normal. You have to consider the DR range of the shot. You exposed for the sky which is boarderline pointing the camera into the sun. The trees and the houses are in the lowest quarter of the sensor range. So basically to expose the houses and trees you could open up 4 stops without burning out.

But to me the image looks perfectly fine and as should, for what it is, you also have flare casing a lowcon glare over the image, that affects the quality of it all.

So to improve it you could HDR-x 2 stops and open up two stops.

Yes the 5k screen makes you see both the details and the error of the images more clear. Possibly you should open up an old r3d and look side by side.
 
I think you are right!
I have to play a little bit more with the camera and compare side by side with old r3d. That´s a very good idea! How the hell didn´t I remembered that! :)

Thanks!
 
Björn Benckert said:
The sharpness is there and the noise levels are normal. You have to consider the DR range of the shot. You exposed for the sky which is boarderline pointing the camera into the sun. The trees and the houses are in the lowest quarter of the sensor range. So basically to expose the houses and trees you could open up 4 stops without burning out.

But to me the image looks perfectly fine and as should, for what it is, you also have flare casing a lowcon glare over the image, that affects the quality of it all.

So to improve it you could HDR-x 2 stops and open up two stops.

Yes the 5k screen makes you see both the details and the error of the images more clear. Possibly you should open up an old r3d and look side by side.
Click to expand...

Hey guys,

I didn´t have time to make a proper test between the two sensors but I picked some random old r3d from the MX and compared them with some from Dragon. I know that this is not a super conclusive test because the light conditions are always different, but as I don´t have the MX anymore it as to be like this. The compressions ratios are pretty close in all the shots, they go from 4K to 6K and I used the same set of lenses. Some have a bit of tweaking in the curves some don´t.
For me the MX look much more clean (Less noise) and sharper. But, I leave a link with the r3d if you guys could help me understand this I would be much appreciated!

Link with the r3d --> Download

Thanks!
 
Guilherme Belchior Afonso said:
Hey guys,

I didn´t have time to make a proper test between the two sensors but I picked some random old r3d from the MX and compared them with some from Dragon. I know that this is not a super conclusive test because the light conditions are always different, but as I don´t have the MX anymore it as to be like this. The compressions ratios are pretty close in all the shots, they go from 4K to 6K and I used the same set of lenses. Some have a bit of tweaking in the curves some don´t.
For me the MX look much more clean (Less noise) and sharper. But, I leave a link with the r3d if you guys could help me understand this I would be much appreciated!

Link with the r3d --> Download

Thanks!
Click to expand...
Guilherme,
thanks for the linkssss .. beautiful places you in Portugal?
Actually you can not compare, each image to be different.
But I see that it is almost always DRagon underexposed at least 1 or 2 and more stops, compared MX movie.
Normal to be a little more noisy.
 
Guilherme, like others have said, the noise level is normal for the way the pictures are exposed which, in most cases for Dragon, is underexposed. Even when the levels are brought back up, I think the images look fine. I do like Dragon's texture, though.

I tend to preach light meters around here (to no avail). An alternative is using Gio Scope to set exposure.

Akin did it to perform his exposure test and I thought it worked quite well.

https://www.youtube.com/watch?v=OmZZjAjJAPQ



Gio Scope displays actual sensor data, independent of metadata adjustments.

Stop 11 represents middle gray. Many people rate Caucasian skin at 1 stop over, which would be 12 on Gio Scope. If you set the key side of your subject's face at 12 and the fill side somewhere in the range from 9-12 (covering a 4:1 ratio), you'll have a consistent frame of reference and target for exposure.

I also noticed you start with REDgamma and then alter the luma curve. Since you're manipulating the curve anyway, might I suggest starting with REDlogFilm instead? I find I get a much better result that way and I'm not trying to undo what the REDgamma curve has already done.
 
Last edited:
You must log in or register to reply here.
Back
Top