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 :)

DEMAND for NEW HIGH ASA CHIP?

Wrote a very detailed post, then safari crashed on my iPod. Sigh.

Basically, you can underexpose and then push a larger sensor more than a smaller sensor, because the added noise is removed by the downrezzing process.

Example (for 2k delivery):

2k 500asa chip has x noise.

4k 500asa chip pushed one stop also has 2*x noise, until we downconvert to 2k, which reduces the noise to x.

I just made those numbers up to illustrate the principle, but I think it makes the point.

Hope that helps...
 
I'm purely guessing, but with Monstro's sensor's wider DR of 13 stops, I'm sure it won't be a 320 ASA sensor. I suppose 800 ASA will be realistic. And if you consider reduced noise though oversampling by downscaling to 1080p 1600 ASA will be a viable option.

That's going to be enough speed.

Hans
 
Hans von Sonntag said:
That's going to be enough speed.

Hans
Click to expand...

The beauty of digital is that it doesnt have the chemical limitations film has. After several generations of cmos, the photography world uses cmos sensors that can crank to 2000 iso and look fantastic blown up, even without noise reduction.

800 would be enough for now.
 
With RED1 all you need is a fast lens.

DR +1 stop will come with Mysterium-X or +2-3 stops with Monstro.

Upgrade your R1 or trade it for EPIC.

But don't worry right now about ASA "faster" chip because you already have enough with 320 ASA with MysteriumR1.

Have a look:

RED_320ASA_REDAlert.jpg

Shot on RED with Canon 50mm FD L f/1.2 @ wide open f/1.2, RED alert output situation.

ASA320.jpg

"Normal" or standard R1 320 ASA output and CC with Orphanage's PanaLog2Lin in Shake that "simulates" Panavision Genesis film out. All is fine here and on a bigger picture it looks better.

ASA400.jpg

400 ASA output and CC withOrphanage's PanaLog2Lin in Shake, already starts clipping highlights.

ASA500.jpg

500 ASA output and CC with Orphanage's PanaLog2Lin in Shake, even more clipping highlights.

All shot on RED1 with Canon 50mm FD L f/1.2 @ wide open f/1.2.

f/1.2 could be something like a T1.3.
 
Crushed blacks are NOT your enemy

Crushed blacks are NOT your enemy

Consider this also,

Most of the fears about low sensitivity steam from the fact that most people seem to believe that they want as much detail in the dark areas as possible, but, do they really? One of the things that makes film so filmic, much more so than dynamic range, depth of field, etc, is the rich, contrasty look it offers, with bold blacks in stark contrast to shiny, light areas everywhere, even in daylight exterior scenes. It is one of the inherent features of film, and one that seems to be overlooked all the time by those tirelessly trying to bring video up to the film level.
Yes, I agree that if you have more dynamic range and a more sensitive sensor, you'll be able to end up with more balanced images, but the truth is that most projects never need that, and truthfully, it is even aesthetically unpleasing. The more neutral your shadows look, the flatter the image you get. Look at most TV shows and films nowadays, most of them shot with huge lighting budgets and on 500 ASA film or sensors, and yet, most blacks are crushed to some degree, because guess what, this gives the image that rich look that most people associate with high production values.
So, if you are going to deliver a product where crushing the blacks will a) reduce or eliminate completely any noise you might get from a slow sensor and b) yield a final product that perfectly complies with the contrasty aesthetics expected, and encouraged, by 99.9% of all productions nowadays, why not stop worrying about a slow sensor that cannot "see" so well in dark scenes and instead use its strengths to their best, i.e. with properly lit scenes, you'll get a beautiful color and tonal rendition that approach that of slide film and easily surpass negative's film? Look at Sanjin's images above, once properly treated, the blacks get crushed a bit, but all noise disappears, the important areas still remain visible, the colors improve dramatically and the image now has that rich, filmic look. I don't need to see into those dark areas, it is a night scene, so I'm happy (and genetically predisposed) to accept that, at night, the sky and unlit areas will look BLACK. Unless of course you are Michael Mann or a proponent of his school of thought...
 
As I understood it, the initial post was about having a 35mm sensor (therefore depth of field) with 2K resolution. and that Tim thought the larger pixels would afford higher EI as a bonus. I think the point is gold. One wouldn't need huge processing power nor take time to down-res etc. and would be able to hit the high-speeds 2K allows for without sacrificing the DEPTH OF FIELD INNIT. that's what it's all about for me.

That and colour space.

If RED made a Scarlet with a 2K 35mm chip, PL mount and did 150fps I'd start saving.
 
Tim Naylor said:
So in essence, are you saying pixel size has little to do with sensitivity/noise?
Click to expand...

Correct. Sensor area and technology have more to do with it, except that, generally, smaller pixels have higher total sensitivity.

Tim Naylor said:
I just found with recent HD cameras they tended to have lower sensitivity when the pixel size was decreased (ie. Varican vs HPX 500).
Click to expand...

My conclusions are based on tests I've done on raw cameras. I don't test video cameras like my XH-A1 because the manufacaturer could be doing a million things in the in-camera software, such as noise reduction. So I can't draw any conclusion about pixel size or sensor size using a video camera: I can only talk about the performance of the entire camera system with certain in-camera settings.

RAW lets you do everything in post-production software, so you can really see what's happening and even compare two cameras on an even playing field.

However, even with that caveat, I'll say that many of the people who prefer SD over HD cameras do so because they are applying a higher standard to the HD camera, like a longer yard stick. They aren't resampling HD down to SD to compare at the same output sizes.

Tim Naylor said:
In any case, so are you saying resolution and sensitivity do not correlate?
Click to expand...

Yes. There are ways that they do relate, though. For example, if every pixel size had the same technology (same QE, read noise, and FWC per area), then as you get smaller and smaller pixels, each pixel has less and less sensitivity per pixel.

Most camera comparisons are done with 100% crops: comparing per-pixel to per-pixel instead of image to image. This leads to misguided conclusions, because the pixel doesn't matter: the final image is what matters. After you resize the pixels down to the same resolution as the large-pixel camera the results are the same (or better).

Tim Naylor said:
If so, what is the major challenge to a higher sensitivity chip?
Click to expand...

EDIT: forgot to answer this question. One of the biggest challenges is physics: there is only so much light to capture, and you can't see light that isn't there. Most sensors already capture between 25-70% of the light falling on them. The theoretical limit is 100%, and after that there's no possibility for improvement, even theoretically. That's why we only see small QE improvements. The D3 had a huge fanfare and it's QE (sensitivity) improvement over the 1Ds Mark III was/is only 15%.

Since sensitivity is so close to the maximum performance, most of the low light performance improvement in the future will come from reduction in read noise, where there is no theoretical limit (but plenty of practical ones!).

Tim Naylor said:
I'm willing to contend with it for now. But I must admit I was let down to see a lot of hype about hyper resolution (28k or whatever) and no mention of increased sensitivity.
Click to expand...

I would expect that is because it's too soon to make comments on the final characteristics of the sensor. However, Jim did spill some of the beans already in this comment:

Jannard said:
Mysterium-X...has 1 full stop more DNR and less noise and is twice as fast as Mysterium
Click to expand...

This could be due to a twice as much sensitivity, half the read noise, or a combination of the two.

Generally people are concerned about sensitivity due to low light performance. Low light performance is often helped by improvement in read noise too, because the shadows and even midtones often fall into the area dominated by read noise. Technically, read noise isn't related to sensitivity, but it is definitely related to low light performance in today's cameras.

We're very close to the theoretical limit of sensitivity (QE), so the only way to get more than a stop improvement will be to use larger sensor sizes, hence FF35, 645, 617.

Tim Naylor said:
The last thing I want to see in 2009 the launch of 5k plus chip and no major advances in ASA.
Click to expand...

[EDIT: fixed quote that was attributed to Jannard instead of Tim.]

Keep in mind that RED's ASA rating is only a recommendation, like every other manufacturer's rating. Even if they make *no* improvement to sensitivity, and all the improvement is only in read noise, then I expect they would increase the recommended ASA. And they should.

Graeme Nattress said:
My light meter was telling me around ISO5000. Tungsten light, B17.
Click to expand...

Wow!
 
Exactly. Low light can look nice on RED. It will only look better as we move forwards, too.

Graeme
 
Thanks for the information Dan,

Cleared up a lot of mysteries of the mysterium. These are the kind of answers I was looking for. Looking forward to the new updates and new chips.

Anyway, it's six days and counting to my first RED gig. I'm armed up to 12k's. Should have more than enough stop. Will need to get more advice on IR's and blue signal but that's another thread.

Thanks all.
 
PS. Dan
The second to last quote "The last thing I want to see in 2009..." you attributed to Jannard, when I think I stated it in a recent post.
 
Well, i can't understand why we should expose REDONE sensor like film anyway.
A sensor does not react like film.

I feel the RED sensor has less highlights headroom over the "usual" greycard exposure.
But i also see that RED sensor has better shadow info.

So i always underexpose (compared to usual greycard method) with the RED, keeping about the same shadow details as film, and thus preserve very good highlight details.
I can lift the mid tones a little without causing banding.

..and it's where raising the mid tones (LUTs in camera) for shooting is so usefull.

This method gave me the superb results.

Antoine
 
Antione, you are using the camera and exposing how you want to, to achieve the results you want!

No digital sensor has any headroom at all. They all clip at a certain brightness levels, so all your range is "below" that clip point. Decide how many stops of headroom you need below clip, and place your mid grey there. Then you can still see down into the shadows below that.

Graeme
 
Graeme,
why can't we use hightlights mask in camera ?
Too much computing or some physical limitations ?
 
Hi Graeme,

That's probably why you said a year ago "expose for highlights".
But i still think that i had to change my way of thinking going from film (photography) to digital sensor.

At least, this conservative method worked great for me.

I prefer to preserve the highlights.
I never found film to have a ton of shadow details anyway.
 
Best that you do that in a colourist package with secondary colour correction where you have control over the process.

Graeme
 
Graeme,

Let's assume i could do a black shading calibration on a perfectly balanced 5% gray surface, could i then be able to use the full 12bit ?

I have not experimented this one...

Just curious... :)

Antoine

EDIT:...because i always have to bring the black level down a little in post anyway...
 
Black shading should be done one black... Lens cap on, iris down.

Graeme
 
Sure sure!!!

...but just for the curiosity...

...would the camera still retain the full 12 bit info even if it would record a little contrastier image ?
 
You must log in or register to reply here.
Back
Top