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 :)

Why not ISO 100?

Yeah, for sure i didn't mean it was only a denoiser, but the noise level has been improved so much...the result is so spectacular...really.

I'm really impressed.
 
Reactor88 said:
Jim,

That is awesome, any chance of seeing the same from the 2 other channels?
Click to expand...

Believe it or not... we really have stuff to do over here.

Jim
 
Thanks Jim! Very useful and exciting information. I hope I didn't open up a chain of "can you do this can you do that". However, I'm glad you posted this.
 
Jannard said:
ISO is only metadata... right up to the point where you change exposure. Then it is not just metadata.

Jim
Click to expand...

If that's not a cryptical note, then I don't know what is...

Could mean that we have a "normal" gain setting in build 16...
 
So all highlights have this very minor difference?

So all highlights have this very minor difference?

Jannard said:
ISO 100, 125, 160- wicked clean image. Image softens as ISO goes down. Need even more light. Use with care due to resolution decrease.

[...] If you lower the ISO to 100, you will reduce resolution and get a "more sterile" looking image. Not much, but it should be noted.

[...] We are not saying the resolution hit is substantial... but it is there due to DSP, mojo, etc.
Click to expand...

Does this mean that the highlights in any RED footage, no matter the exposure or ISO rating, have the same minor decrease in resolution? After all, highlights are just midtones with brighter exposure (i.e. lower ISO).

Jannard said:
[...] Canon has the same issues with their still cameras.
Click to expand...

Does anyone have some images that demonstrate this? I took another look at some low contrast but fine detail in 200, 100, and 50 ISO images from Canon's 5D and 1Dsm2, but did not see it.

(Quotes are from this and the Build 16 ISO thread.)
 
I think I get it now...

I think I get it now...

gunleik said:
If that's not a cryptical note, then I don't know what is...

Could mean that we have a "normal" gain setting in build 16...
Click to expand...

Gunleik:

(Mr. Jannard...correct me if I am wrong here...)

but I think he means that ISO is only metadata until and unless you increase or decrease shutter speed or f-stop on the lens and expose the sensor to light for a longer or shorter period of time.

This makes perfect sense to me now (Thanks!). I guess I just didn't know that sensors are rated like film, I assumed sensors were dynamic and could change their sensitivity to light on the fly anywhere from 100 to 1600 ASA depending on user setting. Now I am curious as to what is the native ISO sensitivity of my Canon Digital Rebel XTi?

Anyway, the other day, I shot ASA 100 primarily because I forgot my ND filter at home, looking back at that image, I do think there should have been a tiny bit more detail and perhaps all of this explains why. I will get the appropriate ND filter and shoot the exact same exterior scene at the same time in the morning at ISO 320 soon and post the difference on this thread. My guess is that it will be nearly unnoticeable . In other words, Mr. Jannard probably didn't have to point this out but it shows a great deal of class that he did.

-shooter
 
Here is the red channel from both. I don't need to post the green channel because they are virtually the same.

Blue channel- biggest improvement.
Red channel- some improvement
Green channel- virtually no improvement (green was pretty good in Build 15).

Jim

9_15d16dred.jpg
 
Hi,

I have a question.

This is how build 16 shot footage will look.

BUT

will these improvements be seen in footage shot on build 15?

David
 
battistella said:
Hi,

I have a question.

This is how build 16 shot footage will look.

BUT

will these improvements be seen in footage shot on build 15?

David
Click to expand...

No... this is why Build 16 is so important. These are firmware improvements.

Jim
 
Let me put on my "know it all" hat and see if I can craft an intelligible explanation.

Let's say (for a really inaccurate, simplistic example) that rating the sensor at ISO 320 means that each pixel on the sensor surface develops an electrical charge of 320 eV for every 320 photons of light that arrive at the the pixel location. Let's make this our 18% grey value. 100% white would then be somewhere near 1800eV. Each stop of light is a doubling of photons or eV charge on the sensor (assuming a linear relationship of photons to charge). Thus an 11 stop range would accept values from 0 to 2047 (thats 2048 steps).

The physics of the sensor cannot be changed (unless we physically replace it), but we can redefine the meaning of the values it delivers and then use some fancy algebra to make the results more usable. By redefining the sensor as ISO 100, we are now saying that 100 photons is 18% grey. This pegs 100% white at 556eV (remember the sensor still has approx 1244 steps available above this value).

Now, instead of having close to 2000 values at hand to describe our various light values between black and white, we're limited to only the bottom 556 values in the sensor's range. That's only a hair over 9 stops - we're throwing away a lot of useful data (although you should be able to recover a lot from your highlights!) and using math to fudge the missing numbers in between.

Jim?
 
Jannard said:
Believe it or not... we really have stuff to do over here.
Click to expand...

Yeah... Like posting in this forum every half hour :tongue:

Seriously, the blue channel improvement is huge.... The noise is much more organic with less hard edges. Even the red channel is noticeable. Can't wait to see it in motion.

Keep up the good work (and get some sleep).
 
So Cool...

So Cool...

Jim,

Thanks for posting the Red channel(I know you guys are really busy), that looks excellent! :w00t:

How does the green compare in this case. Are the Red, Blue and Green channels almost the same noise wise? It looks like Blue is behind Red by only a hair when it comes to noise. I am wondering if the improvement in noise from Red to Green is about the same from Blue to Red?

Also will we be able to deNoise with Red's solution on individual channels?

Just Amazing!
 
Jannard said:
I would shoot ISO 125 with confidence if I was short on ND's, just like I would shoot ISO 1000 if I was short on light. There is less much difference with ISO 125 than ISO 1000. If I had ND's, I'd shoot ISO 320-500.

Jim
Click to expand...

Beyond this resolution hit, I'm looking for good reasons to shoot ISO lower than 320.

If I set up a film camera with 500 ISO Film in it and I shoot a window, exposing for the interior. What's outside the window is just beyond the shoulder of the film, so I loose detail on the window.

If I pop on a different mag with 100 ISO Film, I leave the lense and other parameters the same. The entire image exposes darker, and the window doesn't blow out anymore. I have some detail outside, I loose some detail into the shadows inside.

Ok. So now I set up my Build 16 Red in the same situation. At ISO 500, the window blows out. So I set the ISO to 100. Now everything is darker. But the window is still clipping correct?

So even though my exposure appears to have changed, my main enemies noise in the blacks, and clipping in the whites are in exactly the same place regardless of ISO.

So if I'm correct in my above understanding then I'm having trouble seeing a reason for ever changing ISO metadata on this type of sensor, especially lower.

IBloom
 
The blue channel shot was porno for Redheads. The red channel was, um, Sports Illustrated swimsuit edition.

Having Build 16 loaded into my Red will then be "the real thing". :tongue:
 
Ian,

I'm totally with you.

As I understand it: Build 16 is more 500ASAish than Build 15 was and has less noise in the shadows hence the half stop more sensitivity and considerably more DR.

Overexposing 2 stops (100 ASA) must result in less information in the higlights and more information in the low mids, blacks.

As before all ASA is metadata. Just the way how the sensors data is digitally translated is different and obviously improved.

The algorithms which have to address the white roll-off are vital and must be improved as well if - as Jim stated - shooting 100 ASA is possible without a major sacrify in this regard. Meaning: The dogma "protect your higlights" has change now in a friendly advice.

With a good DI-app I always found the white roll-off of Red pleasing and by far not as bad as some found it.

Hans
 
Hans von Sonntag said:
The algorithms which have to address the white roll-off are vital and must be improved as well if - as Jim stated - shooting 100 ASA is possible without a major sacrify in this regard. Meaning: The dogma "protect your higlights" has change now in a friendly advice.
Click to expand...

I don't think Red is going to implement white rolloff as an algorithm in their camera. That would just mean sacrificing some of the data coming off of their sensor in order to create the appearance of rolloff. The great thing is they give you all the data they have and let you create rolloff with curves in post.

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