Capture One 20 noise reduction is a big step

[DaveRosenthal]

The new noise reduction is really good. At high ISO there is substantially less roughness and better definition of color details without desaturating. Attached is a quick comparison I threw together to show v20 performing at ~1 stop better than v12.

Anyone else have a similar reaction?

[Edit: Oops, too fast. I flipped the labels for v12 and v20--now fixed.]

[kers]

I dont understand exactly:
i do not think you compare : 1600 captureV12    vs     iso 800 captureV20
But that is how i read it...

[DP]

I dont understand exactly:
i do not think you compare : 1600 captureV12    vs     iso 800 captureV20
But that is how i read it...

it is not about the nominal ISO values that he dialed and it is about the exposure (every preschooler knows that ISO is not part of exposure) - dude gave 2 times more exposure (more light hitting the sensor) to the raw used for v20 (  1/30s f6.8 vs 1/60s f6.8  )... granted may be there was a stop worse illumination, but I doubt that - it looks like sequential shots...

so  he illustrated that the old poor C1 v12 can make the raw that was from twice less exposed (1/60) shot to look like the brand new C1 v20 makes raw from the shot that was twice better exposed (1/30)... hilarious !!!

[DaveRosenthal]

it is not about the nominal ISO values that he dialed and it is about the exposure (every preschooler knows that ISO is not part of exposure) - dude gave 2 times more exposure (more light hitting the sensor) to the raw used for v20 (  1/30s f6.8 vs 1/60s f6.8  )... granted may be there was a stop worse illumination, but I doubt that - it looks like sequential shots...

so  he illustrated that the old poor C1 v12 can make the raw that was from twice less exposed (1/60) shot to look like the brand new C1 v20 makes raw from the shot that was twice better exposed (1/30)... hilarious !!!

Oops, mea cupla. I mislabeled the v12/v20 labels in the image--now fixed. So now your comment should read:

"he illustrated that the [new C1 v20] can make the raw that was from twice less exposed (1/60) shot to look like the [old C1 v12] makes raw from the shot that was twice better exposed (1/30)"

The illumination was consistent.

[DaveRosenthal]

I dont understand exactly:
i do not think you compare : 1600 captureV12    vs     iso 800 captureV20
But that is how i read it...

Yes, in my haste I flipped the labels. Now fixed. Sorry all!

The comparison is between (v12 with ISO 800) and (v20 with ISO 1600 and a stop less light via a faster shutter).

[Aram Hăvărneanu]

I much prefer the variants without any noise reduction. Even in this particular example, where the noise is really bad. Noise never really bothered me that much compared to the smooshing of details. Everyone has their own threshold, for sure.

There's no free lunch.

The only time I find noise reduction a net benefit is in astrophotography. The nature of the subject make noise reduction far more effective in that case. You get a free lunch with astrophotography.

[kers]

it is not about the nominal ISO values that he dialed and it is about the exposure (every preschooler knows that ISO is not part of exposure) - dude gave 2 times more exposure (more light hitting the sensor) to the raw used for v20 (  1/30s f6.8 vs 1/60s f6.8  )... granted may be there was a stop worse illumination, but I doubt that - it looks like sequential shots...
so  he illustrated that the old poor C1 v12 can make the raw that was from twice less exposed (1/60) shot to look like the brand new C1 v20 makes raw from the shot that was twice better exposed (1/30)... hilarious !!!

Sorry my problem is i am a post scholar...

I much prefer the variants without any noise reduction. Even in this particular example, where the noise is really bad. Noise never really bothered me that much compared to the smooshing of details. Everyone has their own threshold, for sure.
There's no free lunch.The only time I find noise reduction a net benefit is in astrophotography. The nature of the subject make noise reduction far more effective in that case. You get a free lunch with astrophotography.

I agree, i have usually no problem  with some noise, that is BTW mostly invisible in print.
I have more problems with noise reduction glueing everything together.
In this particular case i think there is a way in beteween...

but anyway, the example you made is clear now.

In the short period i have evaluated C1v20 vs LR classic i see some benefits in the RAW conversion of C1 although LR delivers very clean tiffs with less artefacts.
They are so different that i like to use them both and even mix them together in photoshop for some purposes.
Bu

[ben730]

I much prefer the variants without any noise reduction. Even in this particular example, where the noise is really bad. Noise never really bothered me that much compared to the smooshing of details. Everyone has their own threshold, for sure.

There's no free lunch.

The only time I find noise reduction a net benefit is in astrophotography. The nature of the subject make noise reduction far more effective in that case. You get a free lunch with astrophotography.

+1
But sometimes color noise is ugly....

[rread]

I agree, I hate losing sharpness but when color noise gets too ugly then a little reduction can be helpful. Or just convert to b&w, but that's not always an option.

Thanks for posting this, I haven't had a chance to do any testing yet. When I get a chance, I would also like to do a comparison of the same exposure to see the difference between v20 and v12 with the same file.

[earlybird]

Hi,
 Is some part of the noise reduction incorporated into the RAW conversion before the de-mosaic process or is it a post process made after a raster file has been mapped out?


 Thank you.

[JaapD]

Hi,

Since the noise reduction processing is split up between ‘luminance’ and ‘color’ noise it seems that the processing takes place after de-mosaicing. Before de-mosaicing there is no ‘luminance’.

Regards,
Jaap.

[Bart_van_der_Wolf]

Since the noise reduction processing is split up between ‘luminance’ and ‘color’ noise it seems that the processing takes place after de-mosaicing. Before de-mosaicing there is no ‘luminance’.

Although it is not impossible to let the noise reduction settings have an actual effect on the demosaicing itself (e.g. with a weighted Luminosity of 21.25% Red + 71.54% Green + 7.21% Blue, Bayer filtered photosites), I also think that it operates on the already demosaiced RGB image data. The reason is that it is also possible to denoise TIFFs and JPEGs. Some of the other C1 operations only function on Raw data.

I mostly shoot low ISO images, but on those the Noise reduction is very good and non-destructive.

Cheers,
Bart

[DP]

The reason is that it is also possible to denoise TIFFs and JPEGs.

technically they can consider them as a set of predemosaick "R", "G" (1 instead of 2), "B" channels (once converted from data in containers into some internal representation)... so it is not the ultimate proof

[DP]

Since the noise reduction processing is split up between ‘luminance’ and ‘color’

considering that WB is often given as 2: K & tint sliders one might wrongly assume that it something else but a matrix (3x3) * vector (3x1)... don't make "100%" conclusions just based on how things are presented in the UI

[Bart_van_der_Wolf]

technically they can consider them as a set of predemosaick "R", "G" (1 instead of 2), "B" channels (once converted from data in containers into some internal representation)... so it is not the ultimate proof

Correct, no definitive proof. That's why I said that it's not impossible that it is based on (an RGB weighted representation of) the Raw data itself, and it could therefore also be part of the Demosaicing itself. However, the denoising functionality also supports RGB image sources.

Whatever it is, from what I've seen, it is effective.

[kers]

+1
But sometimes color noise is ugly....

I agree, this is for me the one thing I try to reduce, on high iso i usually do not sharpen at all.

[earlybird]

Thank you for the replies and insights about the technology of the noise reduction.

I am guessing that if Capture One was doing something extraordinary with the technology then the developers would have advertised the fact, and someone here would have known about it, so for the time being I will assume that the noise reduction system is analogous to systems that are available in other post processing environments.

I had asked, because I am primarily interested in using Capture One to process my RAW files and output an intermediary TIFF file for further processing elsewhere, rather than using Capture One as a one step RAW to distribution format processor, and I wondered if the opportunity to reduce noise with a new technology would be lost if overlooked.

Thank you for the helpful information.