DRAGON vs EPIC MX Noise - NOT GOOD

[Robert Ruffo New]

I'm not entirely sure where the "rub" is here, but please look at my post above carefully in terms of white balance, color spaces, and ISO.

Well Phil, I must say your post (helpful, thank you) makes the noise seem about the same on both sensors - worse looking sometimes on Dragon, and worse looking sometimes on MX - this to say that the noise floor seems unmoved between sensors overall.

 

[Meryem Ersoz]

Ruh-roh.

I just received my upgrade last week and have been bashing on the camera quite a bit...shot about 1000gb of footage but haven't processed much of it yet and only shot at 320.

Better take a much closer look at what I've already done and do some comparisons with my MX SCARLET....won't have time until after this weekend, but definitely on my to-do list now...

Will monitor this thread closely.

Thanks for sharing this info, everyone. Hopefully will have some data points to add, when I find a window.

It does make me wonder whether the shipping SCARLET DRAGONS are sharing these issues - would be good to hear from someone who might have ordered one of those, maybe do some comparison shooting....

 

[Robert Ruffo New]

Guys, we are watching this thread. And have noted the people who have brought this to us individually as well. We are investigating internally at the highest levels of engineering. Until we have more data from our tests, premature for us to really comment aside from the fact that "we hear you and are looking into it with all the RED big guns". We will keep the community posted.

And, yes, this thread is what REDUSER is all about.

Kudos to your awesome attitude.

 

[Robert Ruffo New]

I think i will avoid speculating on exactly how they did it, but as the highlights allow 1+ stop more exposure before clip, something in either the optical path or whatever changes sensitivity on the sensor, is PART of what has happened.

As it seems people have non-uniform results, all other things are just stupid speculation. It would not equal the effect similar to change ISO on the old olpf, though

Having shot hundreds of days on MX, the most commonly wanted ISO (meaning the ISO that would have been ideal most often, without needing ND and without having to open up teh lens more than I preferred) was 500 to 800. 250 is just too low - to much wattage needed. The original M used to be 320-250 ish or so and it cost a lot more to light for it. That is to say, in practical terms, I really, really don't want to go back to a 250 ISO camera at this stage - and besides, the old budgets for lights aren't there anymore, and the old OLPF calibration would seem more suited to the day to day realities of shooting, at least in my world.

 

[Robert Ruffo New]

WOW! Looks good enough for moving pictures, to me.

I'm sorry Elise, but some of us have much higher standards than others, and so do our clients. This is not a $2500 camera, and an upgrade should only improve things, not make them worse.

 

[Jarred Land]

The 'OLPF update' was actually a sensor revision

Sorry, showing up a bit late to the party here.

Dunkan, This has nothing to do with the sensor. The new OLPF's improved dynamic range, skin-tones, flares, IR pollution , tungsten performance, as well as a dozen other improvements came about partially by changing the sensitivity in the RED channel ( optically through the OLPF). With the current calibration and the new OLPF the noise is a bit more "colorful" in the gutter than it was on the old OLPF ( and the MX ). Graeme is working on that now via a new calibration and we should see some results next week.

On the flip side as some people already pointed out it is also not fair to the MX sensor to compare it to the Dragon at lower ISO's, because the Dragon at lower than "native" ISO is not at a disadvantage like the M and the MX sensor was , so the Dragon becomes exponentially better the lower in ISO you go.

So lets take a deep breath, let Graeme do his thing and all you guys that are worried can hold off as long as you want... The new OLPF is so much better in so many ways, and the Dragon is better because of it from an image IQ standpoint and it really is the one you really want to be shooting with... the combination of the two together is unbeatable in more ways than not.

Regardless.. at the end of the day.. for whatever reason if you guys really want the old OLPF in your cameras we will be happy to put them back in for you, but it really is taking a step backwards.

 

[Nick Morrison]

Thanks for the update Jarred!

 

[Will Keir]

Same here, I hope this issue can be resolved with updates, I don't want to regret the upgrade.

I am very disappointed with the upgrade and I hope this can be resolved very quickly.

Nice to have you in this conversation, taking for biting into this one. Deep breaths, let Graeme work his miracles, I'm a big hater of "dancing flies," hopefully this is resolved soon. Seems like the solution is to spend more money on hard drive space and shoot at the absolute lowest compression possible to avoid artifacts.

So lets take a deep breath, let Graeme do his thing and all you guys that are worried can hold off as long as you want... The new OLPF is so much better in so many ways, and the Dragon is better because of it from an image IQ standpoint and it really is the one you really want to be shooting with... the combination of the two together is unbeatable in more ways than not.

QUESTION: For use with light meters, what is the sensors rating?

 

[Meryem Ersoz]

we've been here before with RED ONE and MXed EPIC...if it's just a sensor calibration issue, as Jarred seems to indicate, then this is very reassuring.

I think, with the wait we had for RED to "get it right," it seems a little unexpected. but every new sensor has had its firmware tweaked for optimization, after release. If we are just waiting on firmware optimization, I'm actually pretty okay with that...it's just a normal part of being early-ish to the party.... definitely something to test for, and keep an eye on....

 

[Eric Haase]

Sorry, showing up a bit late to the party here.

Dunkan, This has nothing to do with the sensor. The new OLPF's improved dynamic range, skin-tones, flares, IR pollution , tungsten performance, as well as a dozen other improvements came about partially by changing the sensitivity in the RED channel ( optically through the OLPF). With the current calibration and the new OLPF the noise is a bit more "colorful" in the gutter than it was on the old OLPF ( and the MX ). Graeme is working on that now via a new calibration and we should see some results next week.

On the flip side as some people already pointed out it is also not fair to the MX sensor to compare it to the Dragon at lower ISO's, because the Dragon at lower than "native" ISO is not at a disadvantage like the M and the MX sensor was , so the Dragon becomes exponentially better the lower in ISO you go.

So lets take a deep breath, let Graeme do his thing and all you guys that are worried can hold off as long as you want... The new OLPF is so much better in so many ways, and the Dragon is better because of it from an image IQ standpoint and it really is the one you really want to be shooting with... the combination of the two together is unbeatable in more ways than not.

Regardless.. at the end of the day.. for whatever reason if you guys really want the old OLPF in your cameras we will be happy to put them back in for you, but it really is taking a step backwards.

Thanks for posting on this Jarred. CJ Roy posted some grabs of the shots we did and isolated each channel. Just confirming that we really noticed the noise in the red channel big time. I hope that this can be remediated quickly, as the noise in the 0-13 IRE range is really pronounced and a non-starter for any work that I have to shoot at 640 ISO or above. I know I was a huge supporter of fixing the "old OLPF" as I hated that magenta orb flare, but man do I long for clean shadows like we were seeing. I hope that we can have that back again! I know y'all are doing your best on this. If we can help test anything please let us know!

 

[Zeb B]

Thank you for the acknowledgement RED. Looking forward to the fix. Clean 1-20 IRE and color fidelity are important in my work

 

[Matt Ryan]

I was asked to do a comparison between MX at 800ISO and Dragon at 2000ISO, so here it is:

 

[Phil Holland]

Not to divert the conversation here, but I have an odd question that stems mostly from my film background.


What ISO do you want or would expect "crystal clean" ISO performance?


Red's official word on ISO 2000 is "cinema ready" which is relative to common film stocks and grain.


Much like film, Red Dragon's ISO sensitivity effects the general texture of the image.


Which ISO pleases you the most?

 

[Matt Ryan]

Not to divert the conversation here, but I have an odd question that stems mostly from my film background.


What ISO do you want or would expect "crystal clean" ISO performance?


Red's official word on ISO 2000 is "cinema ready" which is relative to common film stocks and grain.


Much like film, Red Dragon's ISO sensitivity effects the general texture of the image.


Which ISO pleases you the most?

I'd expect true 800ISO to be spotless on Dragon which means closer to 1600ISO (Since Dragon is technically a stop under true 800)... It needs to at least be as clean as an Alexa or f55 at comparable ISO.

 

[Eric Haase]

Not to divert the conversation here, but I have an odd question that stems mostly from my film background.


What ISO do you want or would expect "crystal clean" ISO performance?


Red's official word on ISO 2000 is "cinema ready" which is relative to common film stocks and grain.


Much like film, Red Dragon's ISO sensitivity effects the general texture of the image.


Which ISO pleases you the most?

Would love to keep this thread on track with examples of noise and possible solutions from RED as they discover and recalibrate the color science, etc. This is quickly going to turn into a subjective discussion about noise and film grain. This is really about comparing the 2 cameras and not what level is acceptable or pleasing. I think we can all agree we would expect Dragon to be at least as good as the MX, if not somewhat better in terms of noise at any given ISO. I can suggest starting a new thread to entertain everyone's notions about how much noise or film grain they like or dislike in their shots.

 

[Phil Holland]

Would love to keep this thread on track with examples of noise and possible solutions from RED as they discover and recalibrate the color science, etc. This is quickly going to turn into a subjective discussion about noise and film grain. This is really about comparing the 2 cameras and not what level is acceptable or pleasing. I think we can all agree we would expect Dragon to be at least as good as the MX, if not somewhat better in terms of noise at any given ISO. I can suggest starting a new thread to entertain everyone's notions about how much noise or film grain they like or dislike in their shots.

I was actually going to get a bit more technical and not so subjective in terms of measured grain as we have done some pretty great side by sides with some film scans at this point, but okay.


But to focus back in on this a bit and perhaps to speak more of image capture realities. What are you seeing here that is "bad" in comparison to Mysterium-X?

Click image to open in a new tab/window at 100%

 

[Matt Ryan]

What are you seeing here that is "bad" in comparison to Mysterium-X?

It's much noisier than the old olpf/calibration. At the same ISO values the image is darker on Dragon. It appears in several areas to have more noise, also more colorful. I think it's just because everyone expects Dragon to be cleaner considering the old saying "2000ISO is as clean as 800ISOMX. The old calibration pretty much rang true, this one not so much.

 

[Eric Haase]

I was actually going to get a bit more technical and not so subjective in terms of measured grain as we have done some pretty great side by sides with some film scans at this point, but okay.


But to focus back in on this a bit and perhaps to speak more of image capture realities. What are you seeing here that is "bad" in comparison to Mysterium-X?

I'm not looking at the footage the way you have presented it because I would never USE it that way. You have gained it up all the way to 6400 ISO. In an earlier post, I described what is happening as evidenced by the GOALPOSTS and EXPOSURE TOOL when you do this. MX shows more purple areas, Dragon less. MX shows less noise in the Goalpost, Dragon more. When you raise the ISO to 6400 you are trying to get information out of those purple areas. There is a lot less info in the PURPLE areas (which are more present) on the MX so at 6400 yes it (the MX) appears noisier. You have broken the bottom end. Dragon has more RANGE down there, shows less purple areas and doesn't break as bad when pushed to 6400. But when you're not trying to crank info out of nothing Dragon exhibits more noise. Even the camera's meters are telling you it is noisier but has more range in the shadows.

You have also taken the white balance to 1800K, essentially taking out most of the warmth which is where the noise on Dragon is. It's in the red channel and green as well but appears as red/orange noise. You've successfully minimized this by setting WB at 1800. But i would probably never do that for a scene like this. I don't think I've ever set a WB of 1800K on a job.

So just look at the shots in RCX or Resolve at 800 or 2000 ISO with a neutral white balance of 3200 or 4000 or anywhere close to that. This is the most likely way a shot like this would be used. You will clearly see that Dragon has SIGNIFICANTLY more noise than MX. And THAT is precisely what is egregious. Dragon should be cleaner if not at the very least the same.

 

[Phil Holland]

It's much noisier than the old olpf/calibration. At the same ISO values the image is darker on Dragon. It appears in several areas to have more noise, also more colorful. I think it's just because everyone expects Dragon to be cleaner considering the old saying "2000ISO is as clean as 800ISOMX. The old calibration pretty much rang true, this one not so much.

The darker concept is something I'll need to investigate here as I can potentially see what this is about. I would describe Dragon as denser, though middle gray is spot on via the in camera Exposure Tools against my light meter and chart. I'll double check all of this again tomorrow, but we just did all this last week on my camera, so I'm saying it's pretty spot on. There's going to be slight variations here and there versus Mysterium-X due to a variety of reasons though.

The older calibration and OLPF caused artifacts in terms of effected image tonality was a bit off overall. Though a cleaner image possibly due to a lack of contrast, but much more colorful in the shadows compared to the newer OLPF and Calibration. Oddly this would also induce more image noise under certain circumstances where a strong source (or even a high contrast ratio, like shooting silhouettes against a sky) would "nuke" the shadow detail. Old OLPF and Calibration were rather "off" in that way. The new tech produces a far more controlled and consistent image.

 

[Matt Ryan]

Alexas are noisy as hell! I own one.... That said, Been shooting dragon all week, and its scary how clean it is. Perhaps the camera has old OLPF, but still, a 30% noise increase would still be cleaner than my MX, and still cleaner than an Alexa.

You must have an old olpf then... Cause this is the new compared directly with Alexa: http://www.reduser.net/forum/showthread.php?117070-First-Day-with-Dragon