Welcome to our community

Be a part of something great, join today!

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 :)

Build 16 news...

Status
Not open for further replies.
Very Insightfull Mr. Paulsen. I hope your right
 
Yes, this is a great idea, down at the floor level going monochrome will not make a difference. Combaining R and G to fix the B is great idea!
 
So, what I gather is that we won't get beta 16 until after NAB.
 
davecw said:
I'm guessing less noise, more latitude (closer to the stated 11 1/3rd stops...maybe more...who knows:sorcerer: ?) Higher frame rate for 4K (60fps in my dreams, maybe more like 50fps for now? Maybe 3K frame rates will jump too?), 16:9 proxies will finally work, etc? It's already a pretty cool camera, how much cooler can you get?:matrix:
Click to expand...

Let's not get carried away...

Jim
 
Jim "The Voice of Reason" Jannard
 
Blair S. Paulsen said:
1) A softer shoulder on the top end to manage highlights so overexposure looks more like film. This will allow you to open up a bit more to get farther up off the noise floor.
Click to expand...

That would be awesome....although RED already handles the highlights very well...if you handle the RED right. :biggrin:
 
Wait a Min Jim! I sure hope you are not counting Jarred as one of your "Big" things! :P

Ben
 
Bena said:
Wait a Min Jim! I sure hope you are not counting Jarred as one of your "Big" things! :P

Ben
Click to expand...

Nope, I'd thought he counted himself and Mr. Smith and Mr. Smith as the three BIG things :)

Jarred needs to sit before the RED USER to monitor our speculations, meanwhile...

Axel
 
This is an exciting time for us all. I can't wait to see what Red pulls out this year NAB.

I wish I could be there but unfortunately it is not possible.

I am excited to see what build 16 is going to be all about. I could speculate, and say more DR, lowlight, and blue channel fix, I am just happy with what Red has done up till now. Hey it beats the camera I was using hands down, which is no camera.

Keep surprising me Red! I love it!

Good Luck to your future I know it is going to be amazing ride!
 
12-bit A/D vs. DR stops

12-bit A/D vs. DR stops

jbeale said:
Note, given a 12-bit A/D, they can't get more than 12 stops DR without using pre-A/D nonlinearity, which would mean a change in hardware plus a change in philosophy- doesn't seem that likely. Remember that the DR as it exists now is 11+ stops "total" when you disregard noise. Most people do care about noise at some level, so the "effective" or "usable" DR is lower. So there's plenty of room for usable DR improvement without the "total" DR changing at all.
Click to expand...

I hate to be a partykiller regarding the future DR gains Mr. Jannard could achieve but so many seem to think several stops more of DR is simple to achieve.
First a 12-bit A/D doesn't tell you anything about the DR. It could limit the DR if an inferior A/D is used, like a 10-bit A/D on a 10-bit chip, but it will never increase you're DR. A good A/D should be several bits better than the chip to oversample enough to render the lowest stops properly. So if RED uses a 12-bit A/D the sensor is likely giving 9-10-bits max. (real life use) without excessive noise if the designers have done a good job, and they seem to have done that so far by the enthusiasm on this forum.

It is the sensor characteristics which dictates DR. Divide saturation point by noisefloor (photoncount or electroncount) do a log base2 and you get your stops of DR. Or divide the sensor S/N dB by 6 and you also have your Stops of DR - although this is in theory only - as how discerning will the lowermost stop be with only 1 luminance level? Its liek monochrome bitmap, barely doable with a good dither.

My personal experience with MF stills photo CCD's (Leaf Aptus) is you need at least 16 levels (4bits) to render any detail in the deepest shadows without getting posterization and other gunk forming. So the practical max DR with renderd details in all stops on my LEAF back is 7-8 stops (theoretically or the practical pixelpeeper way is 10,5 stops). This sensor uses an 16-bit A/D but DALSA states "only" 11+ bits of usefull info from this sensor and claims its one of the best available. 11+ -4 is 7+ If anyone expects the laws of physics to change by a firmware upgrade they will most likely be disappointed. To get more DR out of the mysterium chip it is probably the de-bayering software, the contrast curve used and filmbased users knowledge that needs an upgrade or to be tweaked, not the camera. Mysterium S/N is about 67dB or was it less? That should relate very well to the 11 stops of theoretical DR - but realworld DR must be lower for the darkest stops to render smoothly above the noise floor. I would expect a live image to render deeper than a still frame as the noise jumps around from frame to frame and lets your eye/brain see through further, so possibly a stop or more could be gained vs. stills. It is strange thought that although the advances in lowering the noisefloor has been tremendous through the last generations of chips (they are now like 9-16 eV), but it is all lost on a similarly decreasing saturation point (100.000 - 60.000eV) due to more megapixels due to the smaller sensel/pixel area which can store the charge.

Maybe the mysterium2 chip will optimize both saturation point and noisefloor to new heights (like 9eV floor and 200.000 eV sat.point with large sensels/pixels would yield 86-87dB S/N and 14+ stops of DR, but if the current S/N numbers is 65/67 the current Mysterium is totally in line with the current cmos chips used in high end dSLR's and will NEVER be able to smoothly render 13 stops as some people seem to hope for. Please consider that an image with 14 stops of DR would look very washed out for more normal scenes and require heavy contrast enhancements clipping to look good.
But keep in mind I am a stills photographer, engineer and consultant - and no RED expert, so everything could in theory run on different physics. But in practice I think a sensor is still a sensor and the laws of physics regarding this application is like it have always been.

Filmbased movie photographers should try to look and learn to the last 5-6 years of advancements in the RAW stills photo workflow and techniques, and learn to be a master in tweaking the RAW footage in post. And getting the new lightmeter from Sekonic which can be calibrated with a custom target and a PC to each ISO/curve on your camera. Then you can measure and shoot on the very limit of your RAW files DR repeatably and with success. Accept there is a fundamentally different aproach to digital RAW captures wheater stills or movie photography.

Note that I consider the RED camera suitable for a lot of what I shoot as a stills photographer and that would give me both stills and clips to sell - I think commercial stills photographers are in for a RED revolution and we will be gaining new territory and probably loose some as well, but life will be more interesting with RED than without. I will wait it out however and wait until late in the year before ordering my RED which will be my new location stills camera 1. and 2. a movie cam...
MHO.
 
Jannard said:
Build 16 is one of the three "small" new things at the RED booth... there will also be three "BIG" new things.

Jim
Click to expand...

3+3 = more than 8 that I can give from HK-CHINA. guess what !! before NAB 2008.

I know my RED answer ......I don't now right from wrong.

Bang WOW Bang
WOW Digitainment
HKG
 
Dare I say a particular camera I work with uses a 32bit fixed point A/D converter and is capable of 13.5 stops which is more then film.....

Build 15 has been a whole new enlightenment as far as what the Red is capable of, so I'm sure 16 will be less about more latitude and more about fps. The Red Channel used to be the quietest channel and now it alot noiser but if thats what it takes to give me 3.5 stops over I'll take it!

DP Art Adams did a great tungsten latitude test with build 15b (2.1.7).

http://provideocoalition.com/index....lesh_tone_latitude_test_under_tungsten_light/
 
Regarding noise performance, has everyone seen this photo?
http://www.reduser.net/forum/showthread.php?t=11353

It's just a test chart, and we don't know the details (lighting color temp, post-processing etc). But have a look at the noise (if you can call it that) in the R,G,B channels.

They haven't said if this is from built 16 (but they didn't say it's not :-) Jarred did aver that it is an actual recorded image frame, and not a CGI render as you might suspect.
 
Dane Brehm said:
Dare I say a particular camera I work with uses a 32bit fixed point A/D converter and is capable of 13.5 stops which is more then film.....
Click to expand...

I'm guessing I'm misunderstanding something, because there is no A/D converter that comes even close to 32 bits of resolution (=> 192 dB dynamic range!). Post-processing signals in 32-bit fixed point is of course possible, but that is unrelated to the resolution of the original A/D, which sets an upper bound on the linear range of your data. The highest I've heard of for video rate A/D is 16 bits, and the operating SNR you get in practice is always lower than just the output word size.
 
I am supersitutes that Build 16 will be released on the 16th April @NAB.


WOW
STEWART
 
Status
Not open for further replies.
Back
Top