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 :)

Double Dragon....

The Rocket has always seemed out of proportion in pricing to the cameras. How is it that my five year old card holds its value better than my 1 year old camera? When the ROI on the debayering card starts beating the ROI on the camera, something is out of whack.

I'm not complaining, because that money is already spent. And I'm very, very happy to hear that my purchase of 2 Rocket cards was not in vain, in terms of the future possibility of the Dragon upgrade. But rather, like many others here, offering it up to the "RED listens" gods as an offering.
 
Jarred Land said:
Hey Filip... Your facts are wrong :)

And if it is really a pain in the ass for you...get a proxy module to record both RAW and super low resolution prores etc. files.. Its not really fair to compare rendering times of 6K raw files vs. 1080p ProRes files.
Click to expand...

Is that a hint? ;)
 
can someone explain to me what it is that makes these shots impressive? :+)

not trolling, just curious what everyone is raving about,
 
Roger Valdes said:
can someone explain to me what it is that makes these shots impressive? :+)

not trolling, just curious what everyone is raving about,
Click to expand...

Shoot that shot with any other camera right now at iso 2000 with that little noise.

It is pretty darn impressive.
 
Roger Valdes said:
can someone explain to me what it is that makes these shots impressive? :+)

not trolling, just curious what everyone is raving about,
Click to expand...

You must be new here. It's the amazingly skin tones and creamy roll off to highlight clip in a 17 stop dynamic range scene, silly. Either that or we so horny for Dragon to ship
 
For some reason, the browser lost all my multi-quote marks. Sigh.

Whenever I ask about CA, nobody answers.

Jarred, ASA might be arbitrary with CMOS sensors but I think you'll agree that it was always the case with negative film (or at least b&w negative). The ASA rating for Kodak 5219 is somewhat arbitrary. A photographer has shown that T-Max (even T-Max 100) can be underexposed by up to six stops and still give a usable image without pushing. Slide film and video are different stories.

I certainly like what I see with the sample images but they look pretty much exactly like the 5K images out of the MX. That is a good thing, considering you've added 20% resolution, extra sensitivity and four more stops of DR.

If anyone thinks that these images look like medium format, they are under a mistaken impression.
 
So what's the Dragon beauty test would be like? Milk Girls 2.0? Or Benedict Cumberbatch and Martin Freeman? (lol i'm joking)
 
This should not be misconstrued to say that RED does not have awesome color science, which it does, but for me, the MOST important factor of any future digital imaging development is the color science. By and large, first and foremost, everything else, resolution, and yes, dynamic range, comes a distant second. Resolution is great, wonderful and impressive, but how many more bloody Ks do we need really? R1 looked darn impressive, still does, and although Epic looks better, I dare say the increase in resolution is not the biggest factor over the R1. Better compression, smaller size et al are far more important. Once we pass 4K I don't really care anymore, I agree with Jim that 1080p was not that great but I am sated with 4K, truly. As for dynamic range, that's funny, most people that peddle the advantages of DR never truly use it. Just like we did with film for so many years, contrast tells the story, and contrast is nothing but good blacks and good whites, and that means (oh God, dare I say it) reduced DR. You can start with a low DR image and need less contrast or you can start with a high DR image and need more contrast, but at the end of the day, most people don't care to find details in clouds or other highlight areas as they don't care to find details in shadow areas. The human brain does not work like that, we automatically discard areas of low detail, be it whites or blacks, and focus on what we can see really well. So yes, it is great to have expanded DR, but at the end of the day, it ain't the panacea most make it out to be. I can use 14 stops, but I am quite happy with 12 stops, and I can even do with 10 stops and get on with the job at hand instead of bitching.

BUT color, well, that's a different story. The ability to extract nuanced tones without incurring noise or artifacting during compression, that is truly priceless. And say what you will, but from an engineering standpoint, a Bayer sensor is not the most color sensitive or accurate solution, so now that RED has managed to achieve resolution beyond belief, insane frame rates and, yes, DR to kick the competition's collective asses, isn't time to make the color science the best there is? In my humble opinion, RED's color rendition and accuracy is not on par with the other aspects I mentioned above. I don't mean the ability to boost saturation to insane levels, I mean, true, nuanced, well balanced, without cross-talk-among-channels color. That is what I wish from RED going forward. Everything else, they have already reached a peak far beyond acceptable...
 
Jarred Land said:
Hey Filip... Your facts are wrong :)

And if it is really a pain in the ass for you...get a proxy module to record both RAW and super low resolution prores etc. files.. Its not really fair to compare rendering times of 6K raw files vs. 1080p ProRes files.
Click to expand...

Facts aside, Filip does have a very valid point. The current RED Rocket card is "old" tech and something faster obviously has to come along to fix the "problem" with 6K. I'm just hoping to see something a bit more price-friendly for the one-man-band kind of guys out there. A $5000 PCI card isn't something everyone can afford and there are incredibly powerful GPU cards out there for 1/10th of the price of a specialized card like the Rocket.

You point aside - how's that Proxy module coming along? (I couldn't wait and bought a Blackmagic Shuttle in the meantime) ;-)
 
Funny how Red's famous (and well deserved) reputation for listening to their customers comes to a screaching halt on the subject of Red Rockets and post decompression. Just sayin'.

BTW, the images from the Dragon look AMAZING.

I wonder how far we can go down the current road though. As pixels get smaller and ISO's get higher, how long before we need to use stops that begin to introduce diffraction issues to take the sharpness edge off, or we need to stack more and more ND's in front of the lens, achieving the same. Isn't it about time for that 6x4.5?
 
Viktor Sloth said:
That sounds quite unbelievable! I wonder how it can do it so fast?
My new beefy Intel i7 system spits out a 3-4 minute video - with effects and all the goodies applied in around 7-10 hours with the CPU at 100 % workload at all time.
Click to expand...

Wow, what kind of quality settings are you rendering out on? I have a 4-5 year old quadcore that renders out 3-4 minute videos with effects etc in 1.5 to 2 hours max at fairly high render settings of 2 passes @ 1080p in Premier
 
One more thing occurs to me. Given the issue of IR contamination, are efforts being made to increase the effectiveness of the IR block filter on the sensor as sensitivity increases? I would say each stop of increased sensor sensitivity should be accompanied by the IR level dropping an extra stop behind the visible spectrum. I mean, no matter what is done technically, sunlight still exposes the same as sunlight always did. Am I making sense?
 
ISO rating it's not arbitrary...
The sensors gets numbers, values.
There is a minimum (dark) value that it can capture and a maximum (white) one.
They are divided by stops. If you can cut in half the light that arrive on the sensor (by closing 1 f/stop, adding 1 stop of ND or double shutter speed) you count a stop.
Starting from quite completely white, cut the light in half and don't stop if you notice something different then the noise floor.
The use of a waveform monitor is advised. Changing ISO is good for expanding shadows/lights since doesn't affect what the sensor can see.

Now, you counted X stops.
Go again at the starting point and stop down X/2 times.
Set your exposimeter to a random ISO value, the current shutter speed taking count if there is any ND filter.
Does it display the same f/stop on your lens? Yes? Ok, that's at what you should rate your camera.
No? Lower the exposimeter ASA setting or raise it and take another measure until it display the same value of your aperture.

Now check that setting this ISO value on the camera in a log format should display the graycard at 44% and at 50% after a video curve.
That's not scientific since the log format has a "storing" gamma and the other one is a "video" gamma.


After all it's your choice to put more stops on the highlights instead of in the shadows (underexposing) just closing by one stop on the lens and raise 1 stop of the ASA. You prefer richer shadows? Do the opposite.

Underexposing or overexposing isn't rate the sensor at another value, because where it clips at 2000ASA clips at 400ASA.


It's also pointless to rate it higher until it reach the same signal/noise of another camera.
ISO is not used to know how much noise I will read in this part of the image. ISO is how much the sensor is sensitive to light.


Case 1: 9 stops over and 9 stop under, rated at 800 ASA, is the same thing as 10+ stops over and 8- stops under, rated at 2000 ASA.
Case 2: 9 stops over and 9 stops under, rated at 2000 ASA, is the same thing as 8- stops over and 10+ stops under, rated at 800 ASA.

Considering that Alexa has 7 over and 7 under at 800 ASA and the Epic MX around 6,5 over and 6,5 under at 800 ASA.

For example you should rate 2 stops over if you want the same latitude of the previous generation on the shadows but with 5 extra stops on the highlights or 2 stops under if you want the same latitude on the hightlight but with 5 extra stop of signal in the shadows.

So, you don't need more ND because it's more sensitive.
It clips further then before.
 
There's no doubt that the new sensor will be great, the lizard looks pretty cool. But until we see a range of images, some skin tones and some high contrast daylight shots I'll temper my praise.

Post processing is a REAL WORLD issue out here. People just love being able to pop footage into their laptop and cut away. PLEASE Red/Jarred/Jim/Ted open up the debayering to the GPU, and release an affordable proxy module at the same time you launch the Dragon. There's no point having the fastest car on the block if you can only drive it on the track.

If Red puts more effort into the workflow then everybody wins. At the moment I'm not sure who is....
 
Hey Jarred!

Glad to see more and more news come out for the Dragon! Really excited to upgrade my Scarlet to it when the time comes.

But for now, I would be more than willing to take a prototype Dragon Epic and give it a run for it's money. Shooting some shorts over the next 2 months for school. After all, I did do an awesome job testing those RedVolt XLs...

Or maybe reverse psychology... no, don't send me an Epic Dragon to play around with... I won't come to NAB to personally hand it off to you...

Anyway, really excited for the Dragon. Everything looks really promising right now. And relieved that I'm keeping my Rocket. Woot! NAB!
 
Rudi Herbert said:
Resolution is great, wonderful and impressive, but how many more bloody Ks do we need really? ..
Click to expand...

The Dragon sensor is the first sensor from Red that will resolve genuine 4K without advanced temporal debayer methods.
I'm sure they are busy implementing the debayer methodology for Dragon, as I tend to think the OLPF will be more 'tight' then the current one on the 5K cameras. By tight, I mean the radius of the effective PSF will be a smaller number of pixels than the old one. Not just scaled proportionality for the new pixel pitch.
 
You must log in or register to reply here.
Back
Top