My Scarlet Dragon arrived - noisy, but otherwise great!

[David Battistella]

Thanks for the explanation Mark. I guess it's time to bust out my old light meter.

Did virtually the exact same test a month ago here. R3ds as well.

http://www.reduser.net/forum/showth...ise-NOT-GOOD&p=1376155&viewfull=1#post1376155

 

[David Kellermann]

Maybe I should clarify that the first image in this thread (the underexposed shot of the girl) wasn't meant to bash the Dragon sensor in general in any way. It's a shot taken at a dinner party with available light, not a film set.

But I was curious why my camera seemed to exhibit much more noise than other cameras (see Jan's photos taken at night in this thread). Here's a new R3D I just shot. This one is definitively not underexposed and not as soft as the others, since I was using the Tokina 100mm macro @ f5.6 instead of the Nikon AIS 50mm @f1.4 as in the other shots.

In my opinion the amount of noise in the midtones is not normal for the Dragon sensor.

window sill R3D

 

[David Battistella]

I's say that is a super clean dragon Frame.

Are you saying there is Color noise in the shadow area of the RED mote between the REC button and the menu wheel. I see that bit of red chroma noise and that is what the firmware is going to address.

Other than that I have to say the Scarlet Dragon blows the EPIC MX sensor away. To me at 800iso that is "normal" and to be expected as things stand at this moment.

If you are going to throw that image out as "too noisy" then I i'd be a bit shocked.

Again here, processed at 250iso REDcolor 4 and Redlog Film I think you have an example of how great this sensor can look. But of course, that is just my opinion. Did Jan post, r3d's? I'm going to check.



Battistella

 

[Jan Reiff]

??

for me, this looks completely ok. what you see at the wall IS the chroma noise, ... this will be fixed. until it´s released, a simple NR will do its job. remember, that RED cameras do nothing in noise reduction in capture process (as many others do)

but i do see nothing, really nothing being an issue here except the known red noise, ... and don´t forget the lense science (Jarred posted this too) - glas can make a difference too.

but honestly, this is all ok, go and shoot with it -

 

[Jan Reiff]

i am not at a fast internet and did not post raw files yet ;-)
but you are right, this file from david kellerman matches well with my footage - and as said, the new firmware and right exposure will hopefully end this discussions ... otherwise i think it´s the wrong camera for some ...

 

[David Kellermann]

I's say that is a super clean dragon Frame.

Are you saying there is Color noise in the shadow area of the RED mote between the REC button and the menu wheel. I see that bit of red chroma noise and that is what the firmware is going to address.

Other than that I have to say the Scarlet Dragon blows the EPIC MX sensor away. To me at 800iso that is "normal" and to be expected as things stand at this moment.

If you are going to throw that image out as "too noisy" then I i'd be a bit shocked.

Again here, processed at 250iso REDcolor 4 and Redlog Film I think you have an example of how great this sensor can look. But of course, that is just my opinion. Did Jan post, r3d's? I'm going to check.



Battistella

If you look at the white window sill board and the darker areas of the white flowerpot, I think there's more noise compared to other RED Dragon shot I've seen.

 

[Jan Reiff]

David (Kellermann)

did you expose at the limit ? it´s is exposed well, but it is close to clipping ? for me it seems there is even bit more room towards the right in the Raw / Redlogfilm - maybe you were able to open even more the lense (which would mean, of course, even more light for the sensor)
i cant´t speak for Scarlet, but the Dragon of the Epic M has a lot of reserves until it´s begins to clip ...

 

[Jan Reiff]

If you look at the white window sill board and the darker areas of the white flowerpot, I think there's more noise compared to other RED Dragon shot I've seen.

not really. i think you really want to see more right now ;-)
we all know, that there is too much red noise with the latest firmware, RED does know and will fix it or did it already as Jarred posted. If the red noise is gone, you will have to look always in 200% crop modus + redlogfilm to see noise in a way, that you may call an issue ...
regarding your camera i would say, some days patience (or weeks who knows) and you don´t want send it back.

 

[David Kellermann]

David (Kellermann)

did you expose at the limit ? it´s is exposed well, but it is close to clipping ? for me it seems there is even bit more room towards the right in the Raw / Redlogfilm - maybe you were able to open even more the lense (which would mean, of course, even more light for the sensor)
i cant´t speak for Scarlet, but the Dragon of the Epic M has a lot of reserves until it´s begins to clip ...

I exposed it to the nearly to the limit. Again I love the way the Dragon sensor handles the highlights. What's worrying me are the midtones / shadows. The point is that there shouldn't be that much noise at 800ASA in the midtones (again see darker areas of the flowerpot or even the dark areas below the window sill).

I have no doubt that RED will fix the chroma noise issue and that the Dragon sensor in general is clean and quite amazing. The point of this thread is to figure out if my Scarlet has a bad sensor, which I think is the case.

 

[Jan Reiff]

when i see the last file of you i think your sensor is okay - again: there is no internal NR process - just put a small NR on it ... use the denoiser in RCX to 6 and crop deep into the record button of your remote in the frame, you will see the difference with just one click - and with the new firmware it will be even better - other cameras do a lot of stuff internal - so we judge always about a naked raw here, ...
honestly (personal taste) this very fine grain gives a very nice kicker to the whole image for me - it´s not a "video" noise what Dragon does (except the chroma noise issue), it´s quite organic. no one wants the red chroma noise, here we are on the same side.
i have Alexa footage here with more noise in same situations (outdoor, on cars for example) - i did test the F-Cameras of Sony, which may be cleaner, but more video-like in the whole image impression ...

 

[David Battistella]

If you look at the white window sill board and the darker areas of the white flowerpot, I think there's more noise compared to other RED Dragon shot I've seen.

Processed at 250 ISO with a WB taken. Full Quality 400%

And 100 percent

 

[Jan Reiff]

go to Redcine, do a 100% view at the record button, use redgamma 4 for example (and dragoncolor of course) - if you slide denoise to the right - you get a much cleaner red button in real time. when the firmware images of Jarred keep what they promise, you won´t even need this, you just need a very light luminance noise coorection, IF you really want or need this super clean image ... fact is, you can get a clean and high dynamic still out of your file without any ADD use - this sensor is okay.

 

[David Battistella]

The point of this thread is to figure out if my Scarlet has a bad sensor, which I think is the case.

I'd say, open a support ticket and file some R3D's and maybe send the camera back, but I have to say that as I can see, what you are seeing is completely normal. I have seen old OLPF dragon stuff look noisier than this stuff you are posting.

Battistella

 

[Laco Zamba]

It's not a issue, it's a feature

 

[David Battistella]

It's not a issue, it's a feature

or or as apple often said,

"it's a feature not a bug"

 

[David Battistella]

Seriously though.

if everyone who posted footage of dragon posted only 100% center crop of dragon in red color 4 REDlog film mode, then you would have seen a lot of this.

If Luigi posted the Alexa Prores frames at 800iso 100percent crop, then I think we'd all be pretty happy.

 

[Ryan Emanuel]

I still don't really understand the complaint. The window sill is white and exposed at 40% IRE. Thats underexposed by at least 4 stops and the noise in the still is really not bad. Every example I have seen has been underexposed. What your calling mid tones are underexposed whites. You can't expect a white object exposed as a dark grey object will be completely unaffected. I would say shoot a test with a regular image with each tone exposed in the region it belongs. If your problem, persists maybe you should send it back, but at least right now it doesn't look like a problem to me.

 

[Ian D]

What is going on with my little lady, I am bit stumped. I used the cam for a wide shot in a simple interview I did yesterday. I thought because I was mixing the footage with s sony 500s that 14:1 compression would do fine. 5k hd, iso 800, canon 24-70 @ 2.8, was pretty well exposed. I find the pic pretty soft which we could put down to the lens but with my pl glass does not make a big difference. When I apply the noise reduction it just shows up red. I'm on the beta firmware, and black balanced accordingly. After a fully debayer out put its just uninspiring.

The second zoom pic is with the noise reduction set to 6, its difficult to see from these but red pixels pop up and the noise does not really decrease.

 

[Matt Ryan]

I still don't really understand the complaint. The window sill is white and exposed at 40% IRE. Thats underexposed by at least 4 stops and the noise in the still is really not bad. Every example I have seen has been underexposed. What your calling mid tones are underexposed whites. You can't expect a white object exposed as a dark grey object will be completely unaffected. I would say shoot a test with a regular image with each tone exposed in the region it belongs. If your problem, persists maybe you should send it back, but at least right now it doesn't look like a problem to me.

Ryan, it doesn't matter whether you expose white at 90IRE or 40IRE, it should still be noise free because it has enough exposure. Yes, the exposure is low and it will be a darker shade of white, but that doesn't mean it's normal to see noise in the whites in this situation. If it was at 10IRE and pushed then of course there should be noise, but not here. There should not be any noise there because it is getting enough of an exposure for this shot. Not every white needs to be exposed so high up on the index. Anything above 30IRE should be pretty clean to be honest, but what I'm seeing is noise currently begins at 40IRE and at times even higher up in the midtones.

 

[Adam Montville]

What is going on with my little lady, I am bit stumped. I used the cam for a wide shot in a simple interview I did yesterday. I thought because I was mixing the footage with s sony 500s that 14:1 compression would do fine. 5k hd, iso 800, canon 24-70 @ 2.8, was pretty well exposed. I find the pic pretty soft which we could put down to the lens but with my pl glass does not make a big difference. When I apply the noise reduction it just shows up red. I'm on the beta firmware, and black balanced accordingly. After a fully debayer out put its just uninspiring.

The second zoom pic is with the noise reduction set to 6, its difficult to see from these but red pixels pop up and the noise does not really decrease.

Sorry, I couldn't resist. lol. But try uploading the pics to an image hosting website. Can't exactly see anything from those screencaps.