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 :)

Iso 4000

Status
Not open for further replies.
I never been around something that just gets better and better each day.
Every technical stuff I ever read about always has some flaws but the RED is just good news... now, if only someone could give me around $25000 so I can begin making my movies, I would in return buy that person a porsche or ferrari when my films start making money... pleeeeease :D
 
:w00t: ISO 4000? Congrats! Can't wait to see some footage...
 
If I shoot ISO 8000 with the dynamic range of RED ONE, I won't even need lighting. So is this noise luminescence or chroma based? I was thinking monochrome would look okay. But CMOS sensors tend to have a lot of chroma noise in my opinion. Also, how much is the dynamic range decreased at ISO 4000 and 8000.

If you reduce the noise at 4K and downsample to 480p You'll get limpidity.

http://www.dvxuser.com/V6/showthread.php?t=72097 Jim talking about ISO 10 months ago
http://www.dvxuser.com/V6/showthread.php?t=75425 Misquoting Jim and talking a bit about vision research

Wow, going back to DVXuser Red fourms... Everything was so quiet there. Unlike REDuser. Even the smallest voice could be heard. And Jarrred wasn't to busy to post.
 
Alexander Nikishin said:
Now imagine some Zeiss Master primes at T1.3 with an ISO of 4,000!
Click to expand...

I'll be using Red with Master Primes in early 2008, and if we can even get close to an ISO of 4000, it's going to give us so much creative flexibility. Of course, rental of the Master Primes costs more than buying the entire Red kit, but that's fair enough.
 
That's why it is called Mysterium,it has got a lot of mysteries yet to deliver,can't wait to see the results:calm: :calm:

Jannard said:
We have successfully done testing at ISO (ASA) 4000. It is noisy but very useable. You might even be able to squeek out ISO 8000. The Mysterium® sensor is so clean that it is amazing.

Jim
Click to expand...
 
Hi,glad to hear that,the same happened to me,watch the results
http://www.flickr.com/photos/7645949@N08/767208684/
http://www.flickr.com/photos/7645949@N08/767208668/in/photostream/
http://www.flickr.com/photos/7645949@N08/767208692/in/photostream/
Shawn Nelson said:
number6, looking back on my own journey, I am so phenomenally glad that 3 years ago I purchased a professional DSLR and learned it well. It has prepared me really well for Red, in ways that I would not have otherwise received by not coming from a film background. A DSLR taught me all about ISO, focal lengths, what the mm in the lenses means, etc. You have several months, I strongly recommend immediately purchasing a DSLR and learning it inside and out. Red is really just a big DSLR that can go at fast continuous FPS.
Click to expand...
 
When I first read the thread heading, I thought Jim was referring to some sort of ISO quality certifications(like ISO:9001, ISO:14000) that company has received.
I am newbee to Cinematography,I had to read every post to realize that you are referring to sensitivity of the sensor.
please don't tell me that Jim actually referring to RED TEN in 2020 :)
 
OK Gaffers,

Sell your generators and invest in LED lighting.

If REDONE can deliver 1000 iso (virtually) noise free, Lord knows what RED2 upgrade will do. You can imagine the demand 5000+ RED owners could put on the development timetable for 24M pixel cmos RED2 sensor, even if it outputs "only" 4K. Noise floor? More like noise basement!

Great news.
 
Jannard said:
We are struggling to assign an ISO rating to the camera. We now think it is 320. But a "trained professional" thinks it is closer to 400-500. Somewhere in there. I guess "officially" (for now) we are stating 320.
Click to expand...

Using an ISO based equivalent for a digital acquisition device is rather subjective as there are no good standards for a testing regiment.

I would suggest a slightly different approach that might yield a more useful specification than ISO. First, establish a system bench mark based upon the device itself. With no applied signal (dark sensor) and the pre quantizer gain set to its minimum value, establish the noise floor in terms of the quantizer's range (say it is +/- 1 LSB). Now, for this same condition establish the illumination level (using any accepted unit of measure) necessary to drive the quantizer to FS. This provides a bench mark of the best case S/N ratio. Next plot the change in dynamic range for increasing values of pre quantizer gain, perhaps in 3 dB steps (or using whatever step size the camera provides). The data pairs (FS illumination / quantizer gain) and the resulting dynamic range each gain step would provide allows someone to fully understand the relationship between gain, illumination level and noise. Not a simple single dimension number like ISO, but a table of the actual properties of the camera under a finite number of conditions. If a single dimension number were really important, this data would give an equivalent input noise specification.

Its an easy test to perform, very repeatable and one that provides highly correlated and useful information. Reminds me of attempting to characterize noise in ultra high performance audio applications, something I do in my line of work. Marketing always wants a single number that they can use in sales literature, engineering always complains that a single dimension spec fails to provide a clear picture of the actual performance and makes comparing one product to the next by published specs alone, a useless endeavor.

The RED is a breakout product and can take a breakout approach to specifications and choose to provide not just the typical ones, but takes an approach which provides real world usable data.
 
For (some kind of vague) reference I dug around for some Canon 1DMk111 shots at iso 3200.
Smugmug
 
KH Muse. Have seen other digital camera makers use ASA type comparisons. Can you explain why this is a bad way to make comparisons? This looks pretty good compared to spec. seen on various newer prof. DSLRcameras which use comparable technology. Does your proposed spec. get used by any of these so we could use it to compare the mysterium to say a Canon Dslr or NIkon Dslr?
For my money this is exciting and means I'm gonna be using smaller wattage lighting a lot!
I'm interested if someone on the forum has ideas about application of noise reduction prior to reducing to HD for output of programming? or 2K for film outs. Will noise reduction at the earlier stage keep more resolution in the final product as I suspect?
 
tj williams said:
Will noise reduction at the earlier stage keep more resolution in the final product as I suspect?
Click to expand...

This may help but only by a small margin. Then again... an aggressive post noise reduction may soften and/or average the image, so yes I guess it would help to noise reduce the 4K first (as you don't want to further soften your 2K or HD down-res.) Again, I think it is wise to really get to know the down-res filtering options to minimize noise (high quality sampling takes much more processing power, is slower, is softer, but can be much better - less zingy.) Also remember that wavelette compression can be thought of as a kind of noise reduction, and is even used in some noise reduction software algorithms - if I'm not mistaken.
 
Hey boothba, can you recommend a post colorist for the LART? Obviously quality post is a big factor in getting a great final image.

Colin, are you sure that that photo from your link was shot at ISO 3200? I have to admit that I'm impressed at the lack of croma in the grain. It actually looks quite nice.
 
Yeah but ASA has never accurately described the response of any recording medium--digital or analog. It's use is in the fact that it's a single number. A single reference point from which to judge everything else.
 
SF Geek said:
Hey boothba, can you recommend a post colorist for the LART testing? Obviously quality post is a big factor in getting a great final image.
Click to expand...

Does it have to be in LA? :tongue: I could hook you up with a top colorist in Toronto - but then you may have to rename the project.... TART!!!!

But back to LA - I've worked with Stefan Sonnenfeld of Company 3, but he'd probably cost an arm and a leg. Perhaps you could contact the folks at Assimilate - although I'm not sure if SCRATCH supports REDCODE yet.

I don't think you need the most extravagant, creative colorist in the world - just someone who understands the technical nature of your tests and therefore doesn't wreck your footage. Sometimes the "famous" guys feel compelled to push your neg (or digital neg) as far as it will go, just to set a new look, justify their rep, or whatever.

It may actually be smart to just use REDCINE so that you don't have to convert hours of footage to uncompressed DPX, TIFFS or CINEON files - which will certainly turn this into a multi-terabyte project. Select shots can then be honed on compositing aps like Shake / Fusion / Nuke / Flame /After Effects / whatever... I'm sure Gavin knows best.

Actually... I could probably pull a favor with Technicolor Burbank or Culver City. Let me make an inquiry.
 
SF Geek said:
Colin, are you sure that that photo from your link was shot at ISO 3200? I have to admit that I'm impressed at the lack of croma in the grain. It actually looks quite nice.
Click to expand...

Well...... I'm sure that the link was specifically about iso3200 shots, but I did not shoot them, so.......

Here is a bigger 3200 example, from a 1D mk3 3200 vs 1Ds mk2 1600 comparison
Rodeo
Says there was chromo noise reduction applied but no luma NR.

I mean, we do not know at the moment, but if this is the kind of image Jim is comparing with...... :w00t:
 
tj williams said:
KH Muse. Have seen other digital camera makers use ASA type comparisons. Can you explain why this is a bad way to make comparisons? This looks pretty good compared to spec. seen on various newer prof. DSLRcameras which use comparable technology. Does your proposed spec. get used by any of these so we could use it to compare the mysterium to say a Canon Dslr or NIkon Dslr?
Click to expand...

I must admit that I haven't seen any DSLR use my suggested measurement approach. My idea was to not be tied to specifications that are to varying degrees, meaningless. I do understand why one would like to have an ISO number to work with (I shoot film so this isn't a foreign concept to me) but the reality is a digital system doesn't lend itself to such a simple, single dimension specification, rather as I mentioned, its a bit more complex. It is the relationship between pre quantizer gain and the resulting noise that is important. Picking a single ISO number is fine but for any given ISO (gain setting) there is a noise characteristic which results, trying to tie it down to one ISO number doesn't reflect the reality of the system. The film equivalent is how a given stock's grain character changes when pushed to varying degrees.

I will certainly perform these tests myself (shouldn't take more than a few hours to complete) and will have the data sets available should anyone be interested in the results.
 
Status
Not open for further replies.
Back
Top