Sensor at high ISO Comparison of cameras.

[Richard Butler]

I don't think we have. The only deletions I'm aware of are those where the original poster has requested that their account be deleted.

Your impression does not appear to comport with empirical evidence. DPR data itself reveals that to date a total 671 (= 17449 - 16778) of my posts have been deleted. Less than around one-third of those deletions occurred in earlier years - a few deletions from what were clearly targeted actions, the rest from wholesale deletions of entire DPR forum threads.

I've spoken to our lead developer, there's been no large-scale deletion of threads. The level of moderation action can change with time, but there's been no change in policy and no intentional removal of threads beyond regular moderation.

The (seemingly evident) more recent disappearances of my posts is around double that. Those numbers sure sound like a lot of alleged (fairly unusual) DPR "self-deportations".

It only takes one prolific thread-starter to ask for their account to be removed for lots of posts to have to be deleted.

I have used Google's site search function for years ("site:URL + SearchTerms"), which is hands-down light years ahead of anything that DPR's home-grown search widget has ever revealed. Around the time that DPR was acquired by present ownership (and since), my similar searches for known keywords in previous posts now yield almost nothing at all.

Google's search algorithms change significantly over time and are a black box. Search terms that used to work now yield no (or, more often, utterly irrelevant) results. That's out of our control.

No change was made to the forum platform and no systematic deletion has occurred since we changed hands.

Richard - DPReview.com

 

[knickerhawk]

While I recall discussing this stuff on older threads to some extent, I'm finding when using Google Site Searches these days that DPR has been steadily deleting tons of older threads.

I don't think we have. The only deletions I'm aware of are those where the original poster has requested that their account be deleted.

With regards ACR, the Camera Raw team told us that the NR and sharpening defaults differ between models, which is why our studio scene processing methodology minimizes both settings and then applies a standard amount of sharpening.

Could you clarify what you mean about minimizing "both" settings and applying a "standard amount" of sharpening since there are multiple sharpening and noise reduction sliders in ACR's Detail panel? What is the "standard" detail-related ACR setting(s) you apply across the board? Thanks

 

[Richard_L]

Your impression does not appear to comport with empirical evidence. DPR data itself reveals that to date a total 672 (= 17450 - 16778) of my posts have been deleted. Less than around one-third of those deletions occurred in earlier years - a few deletions from what were clearly targeted actions, the rest from wholesale deletions of entire DPR forum threads.

The (seemingly evident) more recent disappearances of my posts is around double that. Those numbers sure sound like a lot of alleged (fairly unusual) DPR "self-deportations".

I have used Google's site search function for years ("site:URL + SearchTerms"), which is hands-down light years ahead of anything that DPR's home-grown search widget has ever revealed. Around the time that DPR was acquired by present ownership (and since), my similar searches for known keywords in previous posts now yield almost nothing at all.

(Perhaps) there exist some differences in how much Google takes an interest in storing information pointing to prior DPR Forum posts - or perhaps something related to DPR ?

Hello Detail Man,

Could you give me any of the search terms or keywords that were previously giving you results but are no longer? I'd like to check and make sure Google is effectively indexing our forum content and providing those searches may enable me to locate the posts that you were previously able to find and confirm that they're still available.

Thanks,

(Lead developer) Richard

 

[ThrillaMozilla]

Your impression does not appear to comport with empirical evidence. DPR data itself reveals that to date a total 672 (= 17450 - 16778) of my posts have been deleted. Less than around one-third of those deletions occurred in earlier years - a few deletions from what were clearly targeted actions, the rest from wholesale deletions of entire DPR forum threads.

The (seemingly evident) more recent disappearances of my posts is around double that. Those numbers sure sound like a lot of alleged (fairly unusual) DPR "self-deportations".

I have used Google's site search function for years ("site:URL + SearchTerms"), which is hands-down light years ahead of anything that DPR's home-grown search widget has ever revealed. Around the time that DPR was acquired by present ownership (and since), my similar searches for known keywords in previous posts now yield almost nothing at all.

(Perhaps) there exist some differences in how much Google takes an interest in storing information pointing to prior DPR Forum posts - or perhaps something related to DPR ?

Hello Detail Man,

Could you give me any of the search terms or keywords that were previously giving you results but are no longer? I'd like to check and make sure Google is effectively indexing our forum content and providing those searches may enable me to locate the posts that you were previously able to find and confirm that they're still available.

For what it's worth, the math seems odd, since 17450 is shown as the number of his posts today. This seems rather strange unless the number 16778 refers to the future.

 

[Detail Man]

... the math seems odd, since 17450 is shown as the number of his posts today. This seems rather strange unless the number 16778 refers to the future.

Note: Post count appearing on one's DPR Profile page represents the number surviving (as in following any/all post deletions). The count seen in "post headers" represents all posts.

I had long noted before and during a time of very little posting on DPR that (around) 170 posts had over the years been deleted. That number remained stable for several years.

Today, I find that (around) 400 more of my DPR posts have disappeared. Hand-wavings and sundry declarations do not explain this. I have pretty much given up on my practice of using (very capable) Google Site Search command [site:dpreview.com "SearchTerms"] - because of the small handfuls of DPR posts (using the same various subject related terms as ever) that Google does today find, something around %80 of the hits are "dead" (404). Bum steer city. None of the above evidence supports contention that Google search tools may be to blame.

 

[bclaff]

This seems Off Topic. Maybe you can take it to Site Feedback and Help ?

 

[John Sheehy]

Well, whatever raw viewers should or shouldn't show, they truncate negative numbers, and destroy data in doing so.

You seem to be arguing that a black level should never be set simply because it invariably "truncates negative numbers". Sorry, but that's going too far and is counter to what every raw converter (and raw viewer) does, at least by default. Done optimally, the only "data" eliminated is noise.

Many of the first electron charges from the first photon strikes are discarded by black clipping of the raw data, similar to reciprocity failure, but without the time window aspect. About half of the photosites will be black-clipped, and will waste up to 50% of the first electron charges in the sensor, halving effective QE for very small signals.

When you look at the raw values of a blackframe converted to units of electrons, negative values (when mean black = 0) tell you how many electrons you would need in that photosite before you get the first one (or fraction) which will actually survive black-clipping. Remember that we had sensors a couple of decades back where read noise might be about 30 electrons at base ISO on a 12MP FF sensor; think about how many first electron charges are lost when black-clipping that raw data. The mean sub-black offset value is about -0.8x the standard deviation for a Gaussian distribution, so half the pixels are losing an average of 30 x 0.8 = 24 electrons, or all of the pixels are losing an average of about 12 electrons, with 12 x 12M = 144 million electrons lost, black-clipping before demosaicing. Modern sensors at high ISOs typically have about 1 e- read noise, and such are losing an average of about 0.4 electrons per pixel.

So, ideally, if you really want to avoid discarded electrons, you might want to do everything you're going to do, anyway, that lowers the standard deviation in e- (de-mosaicing, lens corrections, resampling, noise filtering/reduction) before black-clipping.

Getting rid of that is a very good thing indeed because otherwise you're going to be mightly struggling with useless and unwanted "data" (i.e., read noise) and all that this implies, including deep shadow lightening and color casts.

Center-clipping a 1.0e- Gaussian distribution makes the Sigma about 0.584e-, but it is a top-heavy distribution with a mean of about +0.4e- with a milky, veiling appearance with no peek-holes to suggest that black can be under the veil, and the clipping may lose up to 50% of electrons when you need them most, with very small signals. The "benefits" of a lower sigma for a clipped black frame are elusive, IMO. Lowering the readout sigma in an actual image via black-clipping gives you more noise than waiting until going to the RGB display or file to clip, and even there, it is only convention that excludes negative numbers.

 

[JimKasson]

Well, whatever raw viewers should or shouldn't show, they truncate negative numbers, and destroy data in doing so.

You seem to be arguing that a black level should never be set simply because it invariably "truncates negative numbers". Sorry, but that's going too far and is counter to what every raw converter (and raw viewer) does, at least by default. Done optimally, the only "data" eliminated is noise.

Many of the first electron charges from the first photon strikes are discarded by black clipping of the raw data, similar to reciprocity failure, but without the time window aspect. About half of the photosites will be black-clipped, and will waste up to 50% of the first electron charges in the sensor, halving effective QE for very small signals.

So your'e saying that QE varies by number of photons detected, right? Any evidence for that? I can see that at some point it becomes impossible to distinguish read noise from photons detected, but that's a different thing.

Done right, black point correction takes place late in the development process.

When you look at the raw values of a blackframe converted to units of electrons, negative values (when mean black = 0) tell you how many electrons you would need in that photosite before you get the first one (or fraction) which will actually survive black-clipping. Remember that we had sensors a couple of decades back where read noise might be about 30 electrons at base ISO on a 12MP FF sensor; think about how many first electron charges are lost when black-clipping that raw data. The mean sub-black offset value is about -0.8x the standard deviation for a Gaussian distribution, so half the pixels are losing an average of 30 x 0.8 = 24 electrons, or all of the pixels are losing an average of about 12 electrons, with 12 x 12M = 144 million electrons lost, black-clipping before demosaicing. Modern sensors at high ISOs typically have about 1 e- read noise, and such are losing an average of about 0.4 electrons per pixel.

So, ideally, if you really want to avoid discarded electrons, you might want to do everything you're going to do, anyway, that lowers the standard deviation in e- (de-mosaicing, lens corrections, resampling, noise filtering/reduction) before black-clipping.

Getting rid of that is a very good thing indeed because otherwise you're going to be mightly struggling with useless and unwanted "data" (i.e., read noise) and all that this implies, including deep shadow lightening and color casts.

Center-clipping a 1.0e- Gaussian distribution makes the Sigma about 0.584e-, but it is a top-heavy distribution with a mean of about +0.4e- with a milky, veiling appearance with no peek-holes to suggest that black can be under the veil, and the clipping may lose up to 50% of electrons when you need them most, with very small signals. The "benefits" of a lower sigma for a clipped black frame are elusive, IMO. Lowering the readout sigma in an actual image via black-clipping gives you more noise than waiting until going to the RGB display or file to clip, and even there, it is only convention that excludes negative numbers.