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 :)

DRAGON - ITS THE REAL DEAL !!!! - (PART 2)

Chris Henze said:
Any feeling for how long the RedVolt lasts in the side handle compared to the standard Epic?
Click to expand...

I would be prepared for a surprise, but not too anxious. It draws less power itself, generates less heat, has better fans and a new cooling algorithm. Depending upon where you are coming from, you might see a notable improvement. We'll have to wait and see.
 
Last edited:
There's lots to love here - enough to be worth $9500, that's for sure. But- And I say this because I assume constructive comments from users are actually welcome:

1 - On day close-ups, saturation of skin is still anemic. Maybe this will be 100% fixed by new color science - in which cases great. But I do hope it is, at least as an option. What I do prefer about all Sony and Canon cameras is that out of the box, skin is well saturated. No secondaries needed. Maybe this is just a choice Mark made - in which case fair enough.

2 - Mark sometimes says "shot at 250 ISO - but then graded to whatever". This is confusing. What does it matter what you "shot at" if it is still 100% RAW? All you do with a RAW camera is "meter at" or "assume that the sensor is" a certain ISO, what you select as metadata has zero effect on the recorded data. Since it also says it's deliberately underexposed, then Mark was not "metering at" 250 - because then it would be metered-for/ accurately exposed for 250. It makes me question, is ISO still 100% RAW?
 
Mark, amazing as usual. Can I be your cam op? Ill bring my Epic :-)
 
Robert Ruffo New said:
2 - Mark sometimes says "shot at 250 ISO - but then graded to whatever". This is confusing. What does it matter what you "shot at" if it is still 100% RAW? All you do with a RAW camera is "meter at" or "assume that the sensor is" a certain ISO, what you select as metadata has zero effect on the recorded data. Since it also says it's deliberately underexposed, then Mark was not "metering at" 250 - because then it would be metered-for/ accurately exposed for 250. It makes me question, is ISO still 100% RAW?
Click to expand...

ISO was never "raw", it was metadata applied to the raw data capture itself, which was always captured the exact same regardless of the ISO setting (but always completely dependent on the physical exposure settings - shutter, aperture, ND). Conversely, that's also why people still continue to ask "what is RED's native ISO." Whatever it captures at it's base, is the sensors "native ISO" (for lack of a better term) and lowering/brightening it via ISO/FLUT/Exposure/Brightness/Curves/GammaSpace/Etc. is applied to that base raw data -- which was always a FIXED/finite capture.

Dragon seems to be different. It appears that as you adjust ISO, it doesn't just move middle grey like it does on MX, but it actually periscopes it's ~16 stops of DR around that setting (which, to be honest, has always been a better system, as it's basically the exact same as above, but you are no longer "fixed" to that "native ISO" capture.) Other cameras do this via amplification and then bake it in to whatever codec -- RED did the exact same thing, but didn't bake. Dragon does it differently again, only this time it's really not contingent on a native/base ISO of the sensor... Or at least that's how I understand it...

...something tells me I really fubared that explanation.

EDIT: Bare in mind this is all anecdotal, based on that "~16stops (6+/10-?) at 200 and 2000" comment Jarred made.... The post where he said lowering to 100ISO means less ND, which means less IR contamination and glass in front of the lens.
 
Last edited:
Mike P. said:
ISO was never "raw", it was metadata applied to the raw data capture itself, which was always captured the exact same regardless of the ISO setting (but always completely dependent on the physical exposure settings - shutter, aperture, ND). Conversely, that's also why people still continue to ask "what is RED's native ISO." Whatever it captures at it's base, is the sensors "native ISO" (for lack of a better term) and lowering/brightening it via ISO/FLUT/Exposure/Brightness/Curves/GammaSpace/Etc. is applied to that base raw data -- which was always a FIXED/finite capture.

Dragon seems to be different. It appears that as you adjust ISO, it doesn't just move middle grey like it does on MX, but it actually periscopes it's ~16 stops of DR around that setting (which, to be honest, has always been a better system, as it's basically the exact same as above, but you are no longer "fixed" to that "native ISO" capture.) Other cameras do this via amplification and then bake it in to whatever codec -- RED did the exact same thing, but didn't bake. Dragon does it differently again, only this time it's really not contingent on a native/base ISO of the sensor... Or at least that's how I understand it...

...something tells me I really fubared that explanation.

EDIT: Bare in mind this is all anecdotal, based on that "~16stops (6+/10-?) at 200 and 2000" comment Jarred made.... The post where he said lowering to 100ISO means less ND, which means less IR contamination and glass in front of the lens.
Click to expand...

But... Either it's RAW or is' not - if it's RAW there is no reason to mention what you "shot at" because it really doesn't matter - because an image "shot at" 250 ISO and later adjusted to 6400 in RCX will look identical to an image "shot at" 100 000 ISO and adjusted to 6400 in RCX - all you can talk about is what you "metered for" or were "aiming for" in terms of a final ISO in post..

And... if it's a floating "periscope" of 16 stops, as opposed to a flexing gamma curve, but still RAW - that means the recorded dynamic range is much more than 16 stops, because in post I can layer many different ISO exposures in the same frame and composite them with windows. That would be great, if that is the case. Within that, there would still be an ideal sweet spot ISO to choose for metering, with the most latitude - way more than you need, but still - at either end.
 
Are Luke Muggerud said:
I think the reason for the improved compression is that random sensor noise hiding in the dark is reduced with Dragon, and anything that changes on a pixel-level for every frame is hell on compression therefore rendering better results within similar compression ratios with the improved S/N of Dragon
Click to expand...

Correct although RED is intraframe compressed so changes between frames don't matter. That being said, noise does create per-pixel detail that has to be compressed and is murder on compression engines. Then again if you are shooting a naturally noisy subject like a fine textured paper or a forest, or dirt or something else with sub-pixel detail... you're equally going to stress the compression engine with or without noise.

One nice feature for future RED cameras would be an analysis engine which evaluates the uncompressed image for detail and gives you a suggestion for a maximum compression level.
 
Robert Ruffo New said:
But... Either it's RAW or is' not - if it's RAW there is no reason to mention what you "shot at" because it really doesn't matter - because an image "shot at" 250 ISO and later adjusted to 6400 in RCX will look identical to an image "shot at" 100 000 ISO and adjusted to 6400 in RCX - all you can talk about is what you "metered for" or were "aiming for" in terms of a final ISO in post..

And... if it's a floating "periscope" of 16 stops, as opposed to a flexing gamma curve, but still RAW - that means the recorded dynamic range is much more than 16 stops, because in post I can layer many different ISO exposures in the same frame and composite them with windows. That would be great, if that is the case. Within that, there would still be an ideal sweet spot ISO to choose for metering, with the most latitude - way more than you need, but still - at either end.
Click to expand...

Yes, but your "ISO" (monitoring path) dictates your physical exposure settings, which is why pushing something in post wouldn't necessarily look the same. When Mark pushed a ISO250 shot to 4000 (what is that, a 4stop push?), he was basically stretching the shit out of the mid-to-blacks (as it was presumably shot wide-open at ISO250 to get any kind of exposure at all) which means the highlights were being pressed towards clipping (but there's actually still a lot of details in the highs and everything below mid-gray still looked pretty good despite the stretch).

As for the periscope; maybe there is a technical limitation. Maybe they can only fit a certain amount of stops comfortably (with data meaty enough to be manipulation-worthy) within the 16bit REDcode container. The reason they can't go to 32bit is simple; it requires exponentially more processing than Epic has (and would chew up a boatload more data too... aka we're not there yet). Moreover, I think it's a SNR thing - at ISO2000 its comparably clean to ISO200, so you can get away with setting the ISO that high. That being said, a dual-ISO HDRx method (rather than the current dual shutter speed method) would be the cat's PJs and truly offer what you're suggesting (16stops + 5 stops = 21stops total all at 1/48th in two 16bit streams).

...OR maybe what Jarred said is being taken completely out of context (by me and anyone else who read it.)

RED/Jim/Jarred/Graeme/Mark, please jump in at any time to clarify. At the very least, I'd like to be corrected if I'm spewing fud...
 
Last edited:
Mike and Robert, your both correct...

I personally only use the ISO settings to make things brighter for the video village monitors, or if I need to see in the dark, I up the ISO just so I'm able to see what Im doing...
But the fact is, your not changing sensor ISO, your only changing the Metadata settings so you can actually see deeper into what the sensor is capturing. thats all...
So yes Robert your correct, there is and was no need for me to share ISO settings, but in saying that this was the only way for me to show people noise in dark situations, which was to push the META DATA ISO upwards into the 2000, 4000, 6400 ISO range (even though I exposed for a 250iso shot) as this was the only way to do induce noise dramatically.
Silly as it sounds, its was the only way to demonstrate hi ISO range to everyone.

For example : Just say your running in gunning and you chase a man from the bright sunlight as he races into a dark warehouse... He runs from a F22 scene into a f4 scene... Or from a 250iso scene into a 4000iso scene. The question I posed my self was... How much range or how much can I lift in the darks / shadows? As a RAW sensor goes yes... you can almost pull this shot off without even touching your aperture... and this was the reason for that particular test.

As a base ISO goes for the sensor.. and after all our testing and even after much discussion we did not really come to a final conclusion as what that might be...
I thought it might have been between 320-400iso.. My AC felt it was around 400-800Iso... others will think its is between 800iso and 2000iso... Its subjective. At the end of the day, people will park the ISO where they are most happy.

I've decided that I will shoot at mainly 250iso for almost everything, only because I felt the image was smoothest at that point... The highlights are safe if you watch your histogram and this ISO will also save me loading up on the ND's .
Hence why I'm keen as hell for the MOTION mount.
 
Last edited:
steve green said:
Mark,
I'd still love to know the max frame rate (at present) of all the resolutions. Knowing that they may change before release.
Click to expand...

Same as the EPIC MX at this point... I was hoping for more with the test build.
6k WS was 100fps, 6k 2.1 was 87... there was no 6k HD... but Im sure its coming.


Bill Anderson said:
Mark, perhaps I've overlooked this, but what f stops did you use on the car shots:
frames 15 and 24 from the top?
Click to expand...

I was around f4 to 5.6, with some serious ND over the lens to blank out that sun flare.

Mike P. said:
Hey Toia, are you allowed to post a tif/dpx (in RLF)? I'd love to see what any of those 17:1 frames look like at something better than vimeo/jpeg quality. Jim?... Jarred? Bueller?
Click to expand...

This is not the final production ready sensor, It was still a TEST camera, RED are still perfecting everything including the new colour science. So i'm expecting even better image performance than what I shared with you all. So when its 100% ready to go out, I'm sure they will load up a proper 6k image for you all to scrutinise over. :)

Robert Ruffo New said:
There's lots to love here - enough to be worth $9500, that's for sure. But- And I say this because I assume constructive comments from users are actually welcome:
On day close-ups, saturation of skin is still anemic. Maybe this will be 100% fixed by new color science - in which cases great. But I do hope it is, at least as an option. What I do prefer about all Sony and Canon cameras is that out of the box, skin is well saturated. No secondaries needed. Maybe this is just a choice Mark made - in which case fair enough.
Click to expand...

Uuum... not sure how to respond, but maybe my daughter looked Anaemic. The image / color / looked exactly colour perfect to actual human being standing in front of me. I'll tell her to get some more sun. :) Maybe your monitor your watching the Compressed quicktime on is a little flat? So many variables.
Looking at this shot via a RETINA Ipad is the closest thing to perfect... I'm looking at it now with my daughter sitting beside me, looks exactly the same.
Sorry Rob, your not going to get a bite out of me.

paul schefz said:
glad this one is getting a fresh start....

still wondering about compression and workflow...
17:1 seems unbelievable to me, i try to stay at 6:1 with my epic...
also: you mentioned somewhere that you were able to play back full 6k in RCX at 1/4 on your mbp?!?!
Click to expand...

6k playback via RCX at 1/4 res ... yes.... Via my MBP. no Rocket inline.
17:1 was clean... Seemed unbelievable to me too... but its fine.

paul schefz said:
it somehow does not make much sense that we are getting more pixels with more DR at a much more flexible iso range and get smaller files that are easier to work with? don't get me wrong....i am all for it....any input on this?
Click to expand...

I have no idea how they do it.

Chris Henze said:
Any feeling for how long the RedVolt lasts in the side handle compared to the standard Epic?
Click to expand...

We were getting an extra 3 to 5mins out of them with the DRAGON.
 
Last edited:
Mark Toia said:
I've decided that I will shoot at mainly 250iso for almost everything, only because I felt the image was smoothest at that point... The highlights are safe if you watch your histogram and this ISO will also save me loading up on the ND's .
Hence why I'm keen as hell for the MOTION mount.
Click to expand...

Interesting. I'm real curious about shooting at lower ISOs as well. I have a feeling I'll expose/light for midtones with ISO 500-1000 in mind to give me more "meat" in the grade. Eager to test out and find where that sweet spot is for the fattest possible "digital negative". We'll see where that actually lands in regards to the histogram and what Raw sees.

I like to explore some not so typical grades in post and enjoy pushing those boundaries, stretching values and moving color around a lot. Mysterium-X w/ REDCODE is fairly liberating on that front already. Curious where Dragon brings this to the next level is for sure.

ISO 2000 and I'd say up to around ISO 4000 are certainly going to be usable straight out of camera from what I'm seeing. Maybe even up to ISO 6400 based on your tests here Mark because that boat in the water looks rather lovely to my eyes.
 
Phil Holland said:
Interesting. I'm real curious about shooting at lower ISOs as well. I have a feeling I'll expose/light for midtones with ISO 500-1000 in mind to give me more "meat" in the grade. Eager to test out and find where that sweet spot is for the fattest possible "digital negative". We'll see where that actually lands in regards to the histogram and what Raw sees.

I like to explore some not so typical grades in post and enjoy pushing those boundaries, stretching values and moving color around a lot. Mysterium-X w/ REDCODE is fairly liberating on that front already. Curious where Dragon brings this to the next level is for sure.

ISO 2000 and I'd say up to around ISO 4000 are certainly going to be usable straight out of camera from what I'm seeing. Maybe even up to ISO 6400 based on your tests here Mark because that boat in the water looks rather lovely to my eyes.
Click to expand...

Yes that 6400 boat shot did catch me by surprise as well.... And yes, 2000 and 4000 are completely usable straight out of the box.

Everyone is going to find there own sweet spot with the dragon.... and yours Phil, could be completely different to mine. :) ...
 
It makes sense.

With the lower noise floor you can easily push ISO higher in post because as you move up the "gain" there is much less noise being pronounced. No noise at the onset means no noise when you push the ISO in RCX.

Since I am a 320 ISO guy on MX I trust your rating Mark at 250 ISO, that is exciting.

David
 
Mark Toia said:
Mike and Robert, your both correct...

I personally only use the ISO settings to make things brighter for the video village monitors, or if I need to see in the dark, I up the ISO just so I'm able to see what Im doing...
But the fact is, your not changing sensor ISO, your only changing the Metadata settings so you can actually see deeper into what the sensor is capturing. thats all...
So yes Robert your correct, there is and was no need for me to share ISO settings, but in saying that this was the only way for me to show people noise in dark situations, which was to push the META DATA ISO upwards into the 2000, 4000, 6400 ISO range (even though I exposed for a 250iso shot) as this was the only way to do induce noise dramatically.
Silly as it sounds, its was the only way to demonstrate hi ISO range to everyone.

For example : Just say your running in gunning and you chase a man from the bright sunlight as he races into a dark warehouse... He runs from a F22 scene into a f4 scene... Or from a 250iso scene into a 4000iso scene. The question I posed my self was... How much range or how much can I lift in the darks / shadows? As a RAW sensor goes yes... you can almost pull this shot off without even touching your aperture... and this was the reason for that particular test.

As a base ISO goes for the sensor.. and after all our testing and even after much discussion we did not really come to a final conclusion as what that might be...
I thought it might have been between 320-400iso.. My AC felt it was around 400-800Iso... others will think its is between 800iso and 2000iso... Its subjective. At the end of the day, people will park the ISO where they are most happy.

I've decided that I will shoot at mainly 250iso for almost everything, only because I felt the image was smoothest at that point... The highlights are safe if you watch your histogram and this ISO will also save me loading up on the ND's .
Hence why I'm keen as hell for the MOTION mount.
Click to expand...

Mark - I meant no offense to your daughter. I felt that both the man and your daughter looked a little flat and a little off - still what that slightly excessive yellow I always grade away with secondaries, but wish I didn't have to (although it's already better than uncorrected MX)- it has nothing to do with their actual skin tones, as they both looked off in the same way. To my credit, I have something a little fancier than a retina display here, and much more precise, calibrated every two weeks to within 0.3% of Red 709 perfection using a Hubble probe that is itself calibrated every 6 months, with perfect 5600K backlighting on a wall panel painted with special neutral grey paint. It's not my monitors/projectors leading me astray. The problem really is there, although it's subtle.
 
Robert Ruffo New said:
Mark - I meant no offense to your daughter. I felt that both the man and your daughter looked a little flat and a little off - still what that slightly excessive yellow I always grade away with secondaries, but wish I didn't have to (although it's already better than uncorrected MX)- it has nothing to do with their actual skin tones, as they both looked off in the same way. To my credit, I have something a little fancier than a retina display here, and much more precise, calibrated every two weeks to within 0.3% of Red 709 perfection using a Hubble probe that is itself calibrated every 6 months, with perfect 5600K backlighting on a wall panel painted with special neutral grey paint. It's not my monitors/projectors leading me astray. The problem really is there, although it's subtle.
Click to expand...

But still, you would agree that this came off pretty well for quick and dirty test footage, no? I'm pretty sure you are not suggesting Mark can't get a proper skin tone, so I am assuming you are not laying the blame for the "bad skin" on Mark, correct?

Rob, I think you are like those people who are known as a "Nose" for sniffing perfume or a "Palette" for tasting wine or cheese. I think you have proved that you have brain/eye coordination that well exceeds the norm.

Thanks for pointing out these flaws that the rest of us cannot see.
 
Last edited:
Rob, with regards to the anaemic skin tones, I'm seeing the same thing you are.

Its due to a too weak curve being applied to REDlogFilm footage, which in turn isn't accurately placing the log values where they need to be to properly de-log REDlogFilm. This is most apparent on well modelled skin as it occupies a large slice of the log range. If the curve isn't right, a large portion of the skin is left quite flat and most of it will be left rather desaturated. Apply the correct curve, or even just a slightly sharper curve and not only will the values be closer to where they need to be, but the sharper curve will also bump the image saturation. Both of these lead to more traditionally pleasing skin tones.

Mark mentioned this exactly, stating that he took the footage into RCX for "a very basic curve which was applied to a RED LOG film setting".

This isn't really a wrong way of doing it, in fact it's probably the best way to show the entire range captured by the camera. As a basic, less aggressive curve means there is limited compression to the information in the shadows and highlights, which better shows off the entire captured range. After all, showing the entire range of the sensor was the main goal behind the stress test, correct?

This is actually the workflow used in a lot of low-con, desaturated grades on TVCs etc. As its easier to create your own custom curve for log gamma images that places the values and contrast where you want it. Rather than starting from a standard log2video curve and trying to work back from there.

I'm happy to put up some examples that illustrate what I'm talking about...
 
Elsie N said:
But still, you would agree that this came off pretty well for quick and dirty test footage, no? I'm pretty sure you are not suggesting Mark can't get a proper skin tone, so I am assuming you are not laying the blame for the "bad skin" on Mark, correct?

Rob, I think you are like those people who are known as a "Nose" for sniffing perfume or a "Palette" for tasting wine or cheese. I think you have proved that you have brain/eye coordination that well exceeds the norm.

Thanks for pointing out these flaws that the rest of us cannot see.
Click to expand...

No, I saw it too. I have dealt with this by curving the LUT in my own Scarlet in such a way that generates better color response. But there is always a trade off to be made between smaller photo sites/resolution vs larger photo sites/color. While one may be good enough for some, the other may be not enough for others. Pick your poison.
 
you mean for only $9500, I get to shoot images as beautifully as Mark Toia? such a BARGAIN!!!!

oh wait....that's not how it works?

I'm still in awe of how much ground you covered in two days. still beautiful. still excited for the new sensor.

I actually was on the fence, because I have no complaints, none whatsoever about MX, there is already an astonishing amount which can be pulled from the RAW, if it's needed. I've been far more interested in waiting on some of these support items, modules, etc., which can streamline the camera and enhance its functionality, than in throwing down for a new camera.

The Toia test has changed my mind. I'm way more excited to throw down for this new camera now that we've seen a glimpse (just a glimpse, as per Jim's clarifications) of what it can do.

Keep playing with your knobs, lads! It will only get better from here.

snicker....
 
You must log in or register to reply here.
Back
Top