PIX 2015
NarrBL

NarrBL

Lives in US Minor Outlying Islands US Minor Outlying Islands
Joined on Jun 6, 2003

Comments

Total: 115, showing: 1 – 20
« First‹ Previous12345Next ›Last »
In reply to:

NarrBL: You know, I read this, and then the original article, particularly some very interesting comments from Ilya. Then I think we aren't understanding this compression very well, and that indeed the Sony marketing statements may be correct, in an important way.

First, Ilya calculates that there isn't a problem with the compression, but rather an error in the decompression stage of the raw processing. If correct, it's something that could be remedied in firmware, I think, and would then eliminate the 'stripe' artifacts.

But the second point comes from thinking about how the compression is explained to work -- and this is where important dynamic range, particularly the microcontrast and shadow detail, actually can be better with Sony's scheme.

Consider the sharp jump in intensity - like the uncratered right side of the moon capture. Yes, the algorithm is going to take the bright and dark intensities, then use less than full bit-depth steps to record the transition between them.

(cont'd)

Anyway, while I was writing this, ProfHankD posted above about his analysis and repair approach, which may support Ilya's thinking - we'll have to see.

In any case, good to have someone going further than my intuitions here, and I note he still suggests that Sony offer both the uncompressed and compressed algorithms.

We might then really be able to see to what degree the compression may help microcontrast and shadows, which would be pretty interesting.

This is a territory Sigma has been very good on, within the ISO limits of their own sensor approach, and those images do look very good..

Direct link | Posted on Sep 3, 2015 at 04:01 UTC
In reply to:

NarrBL: You know, I read this, and then the original article, particularly some very interesting comments from Ilya. Then I think we aren't understanding this compression very well, and that indeed the Sony marketing statements may be correct, in an important way.

First, Ilya calculates that there isn't a problem with the compression, but rather an error in the decompression stage of the raw processing. If correct, it's something that could be remedied in firmware, I think, and would then eliminate the 'stripe' artifacts.

But the second point comes from thinking about how the compression is explained to work -- and this is where important dynamic range, particularly the microcontrast and shadow detail, actually can be better with Sony's scheme.

Consider the sharp jump in intensity - like the uncratered right side of the moon capture. Yes, the algorithm is going to take the bright and dark intensities, then use less than full bit-depth steps to record the transition between them.

(cont'd)

But since the transition occurs over very few pixels in image size, we will not see any great inaccuracy. Even if it were recorded at 14 bits, we would not see much detail, because there isn't much detail.

Now consider where we so often notice issues, in microcontrast over a smooth surface, or in 'blocking' of dim detail of shadows.

Here the same algorithm will act to give more bit depth for this portion of the image than a non-compressed raw would give. It is over some area that it decides marking the high and low intensity, then using its full resolution between these high and low points. If there is noise that would be visible, it is taking quite accurate measure of this on each pixel, minimizing the false transitions noise would cause, for example the so-called 'blocking', also allowing a good noise filter to do its best work.

To degree I may be correct in this thinking, it appears such a dynamic compression algorithm may be a very effective ability for a high end camera.

Direct link | Posted on Sep 3, 2015 at 03:51 UTC

You know, I read this, and then the original article, particularly some very interesting comments from Ilya. Then I think we aren't understanding this compression very well, and that indeed the Sony marketing statements may be correct, in an important way.

First, Ilya calculates that there isn't a problem with the compression, but rather an error in the decompression stage of the raw processing. If correct, it's something that could be remedied in firmware, I think, and would then eliminate the 'stripe' artifacts.

But the second point comes from thinking about how the compression is explained to work -- and this is where important dynamic range, particularly the microcontrast and shadow detail, actually can be better with Sony's scheme.

Consider the sharp jump in intensity - like the uncratered right side of the moon capture. Yes, the algorithm is going to take the bright and dark intensities, then use less than full bit-depth steps to record the transition between them.

(cont'd)

Direct link | Posted on Sep 3, 2015 at 03:45 UTC as 39th comment | 2 replies
On photo in sample gallery (1 comment in total)

this one quite artistic...thanks ;)

Direct link | Posted on Sep 1, 2015 at 22:59 UTC as 1st comment
On photo in sample gallery (1 comment in total)

a photo of great beauty...

Direct link | Posted on Sep 1, 2015 at 22:57 UTC as 1st comment
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

Richard Schumer: I am very disappointed with the IQ of the images posted in the gallery. I dl'd the DNG files, converted them to tiffs and messed around with one (the fisherperson) according to my preferences. The shadows can be pulled up quite a lot, and there's much information in the highlights, too, but the shadows are noisy with chrominance artifacts. Overall noise at ISO 800 seems like my old Pentax 10d -- not up to current standards, IMHO based on my Olympus E-PM2 at the same ISO.

But, there is objectionable IMO vignetting at f:2.2, made worse by my usual processing. See the conversion in my gallery -- js_ISO800_f_2.2_L1010969
(I do not know how to post the photo in this forum -- sorry. If someone would like to enlighten me....)

AND -- if one looks at the waterline, there appears to be a huge amount of barrel distortion, as well, considering how high up in the frame the horizon is here.

Richard, I think there's a misunderstanding here.

The raw files are not 'doctored'. That's why you can see the image before correction, when you use a raw tool that isn't properly set up for modern cameras.

Seeing that doesn't show that anything is wrong or less than professional in a camera's design, or its clientele, certainly.

I think there's something missed in thinking about photographers in such an old-fashioned classifying way also, but that's my opinion. Good fortune with your photography.

Direct link | Posted on Sep 1, 2015 at 22:09 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

Richard Schumer: I am very disappointed with the IQ of the images posted in the gallery. I dl'd the DNG files, converted them to tiffs and messed around with one (the fisherperson) according to my preferences. The shadows can be pulled up quite a lot, and there's much information in the highlights, too, but the shadows are noisy with chrominance artifacts. Overall noise at ISO 800 seems like my old Pentax 10d -- not up to current standards, IMHO based on my Olympus E-PM2 at the same ISO.

But, there is objectionable IMO vignetting at f:2.2, made worse by my usual processing. See the conversion in my gallery -- js_ISO800_f_2.2_L1010969
(I do not know how to post the photo in this forum -- sorry. If someone would like to enlighten me....)

AND -- if one looks at the waterline, there appears to be a huge amount of barrel distortion, as well, considering how high up in the frame the horizon is here.

Hi Richard, yes, and I posted also in the sense of community, but also on your interests. I think we have a good opportunity in being interested, without needing to be purchasing the camera.

I think the thing to realize is that it's this less compensated lens combined with advancing digital correction that has been making possible so many of the smaller cameras with high lens performance that have appeared, whoever uses them. By this I mean the RX100s, the mirrorless that feel a real advance, and so forth.

It's true that there can be a degree of issue in corners from this, but this is where the careful design intention comes in with new tools, just as used to involve only glass optics.

I know it's something I looked at carefully when choosing a small camera, and saw differences. The Olympus XZ-2 I settled on is particularly good on this, so that there is no visible problem, and that''s what I'd expect from a Leica design of this type as well - along with excellent color, etc..

Direct link | Posted on Sep 1, 2015 at 19:33 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

dkov70: HELP! Is anything wrong with the new 100% viewing widget on dpreview or with my iPad screen? All images, also at 100 ISO look at 100% magnification soft and blurry.

Yes, the galleries are still broken on my iPad also, I think from something changed over the weekend.

It has been mentioned, and I'm sure the DPReview web persons are looking to fix it.

Direct link | Posted on Sep 1, 2015 at 18:57 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

Richard Schumer: I am very disappointed with the IQ of the images posted in the gallery. I dl'd the DNG files, converted them to tiffs and messed around with one (the fisherperson) according to my preferences. The shadows can be pulled up quite a lot, and there's much information in the highlights, too, but the shadows are noisy with chrominance artifacts. Overall noise at ISO 800 seems like my old Pentax 10d -- not up to current standards, IMHO based on my Olympus E-PM2 at the same ISO.

But, there is objectionable IMO vignetting at f:2.2, made worse by my usual processing. See the conversion in my gallery -- js_ISO800_f_2.2_L1010969
(I do not know how to post the photo in this forum -- sorry. If someone would like to enlighten me....)

AND -- if one looks at the waterline, there appears to be a huge amount of barrel distortion, as well, considering how high up in the frame the horizon is here.

(...continued...)

If you are want to look further into it, here are some links which seemed to give good hints:

http://www.darktable.org/2012/10/whats-involved-with-adding-support-for-new-cameras/
http://photo.stackexchange.com/questions/61791/how-do-i-add-lens-correction-for-my-modellens-to-lensfun-in-darktable/61793#61793
http://sourceforge.net/p/darktable/mailman/message/34141817/

An alternative thing you could do is download a 30-day free trial of Photoshop, if you want to seriously play with the images DPReview is providing, and if you have recent enough Windows or a Mac available, as it sounds like you are using Linux.

Direct link | Posted on Sep 1, 2015 at 06:38 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

Richard Schumer: I am very disappointed with the IQ of the images posted in the gallery. I dl'd the DNG files, converted them to tiffs and messed around with one (the fisherperson) according to my preferences. The shadows can be pulled up quite a lot, and there's much information in the highlights, too, but the shadows are noisy with chrominance artifacts. Overall noise at ISO 800 seems like my old Pentax 10d -- not up to current standards, IMHO based on my Olympus E-PM2 at the same ISO.

But, there is objectionable IMO vignetting at f:2.2, made worse by my usual processing. See the conversion in my gallery -- js_ISO800_f_2.2_L1010969
(I do not know how to post the photo in this forum -- sorry. If someone would like to enlighten me....)

AND -- if one looks at the waterline, there appears to be a huge amount of barrel distortion, as well, considering how high up in the frame the horizon is here.

Ok, I looked into this for you -- to a degree sensible.

I think darktable ought to handle the lens correction for you. It has such abilities, as these would be necessary for cameras since about 2012 it appears.

How it does this seems to be by recognizing the lens identity from the image's EXIF information. The name must be an exact match to the lens contained in (apparently) the Lensfun database. Others have duplicated information from a differently named version of the same lens.

Here is probably a problem, as the Leica Q's lens sounds unique, and likelyno-one has calibrated it in Lensfun.

What you could do is manually correct a shot like the one you tried, and then use the settings for other similar shots, doing the same for each f/stop range you are interested in.

Direct link | Posted on Sep 1, 2015 at 06:36 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

Richard Schumer: I am very disappointed with the IQ of the images posted in the gallery. I dl'd the DNG files, converted them to tiffs and messed around with one (the fisherperson) according to my preferences. The shadows can be pulled up quite a lot, and there's much information in the highlights, too, but the shadows are noisy with chrominance artifacts. Overall noise at ISO 800 seems like my old Pentax 10d -- not up to current standards, IMHO based on my Olympus E-PM2 at the same ISO.

But, there is objectionable IMO vignetting at f:2.2, made worse by my usual processing. See the conversion in my gallery -- js_ISO800_f_2.2_L1010969
(I do not know how to post the photo in this forum -- sorry. If someone would like to enlighten me....)

AND -- if one looks at the waterline, there appears to be a huge amount of barrel distortion, as well, considering how high up in the frame the horizon is here.

Richard, I've just tried that same DNG, and actually don't see visibly or measure any of the vignetting or distorting curvature you show in your gallery image.

I wonder what Photoshop you are using? It has to be new enough to be able to read the corrections embedded in the DNG file, as we've mentioned is normal for fixed-lens cameras (and some detachable) these days. I have CS-6, not very new, but with the last (9.1) Adobe Camera Raw for it, which does indicate it's using the corrections.

Capture One 8 does the entire image even better, also using the embedded corrections, and there as well as in Photoshop, the noise appears to be mostly luminance, so the chrominance noise you see may also be a matter of the development software.

Both of these softwares think the original is underexposed by 0.5 to 1 f/stop, but you can still get the face to show quite reasonably even though it is very dark indeed beneath that anti-glare hat bill.

So, a tough shot, but the camera's got it, I think.

Direct link | Posted on Sep 1, 2015 at 00:14 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

mononk: I certainly do not appreciate the type of pictures you post on these so called "real world sample gallery". Would you publish any of these shots in your SHOW CASE. Come on guys, work a little harder, you are supposed to be pros.

You know, what an artist does is translate, turn what tools can do and an eye can see into a result which is evocative.

If you want to see things already done, perhaps you want DeviantArt, not as deviant as it tries to sound.

Here we're interested in what _can_ be done, and seeing natural photos taken in natural conditions is a very good way.

That our correspondents like to show the life around them is a privilege -- and grounds the sense of what you are really going to get with a camera or a lens. At least as I see it...

Direct link | Posted on Aug 31, 2015 at 21:47 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

User9249539467: Up until the time that full frame digital cameras fell below the $3000 mark, there were many different reasons to choose a specific brand. My preference was Nikon because they were the only ones that paid attention to the color characteristics throughout their catalog; they all matched, and I shot mostly positive films. My goodness, Leica lenses were wonderful, but each lens was different. Now... we even have Sigma producing decent lenses! And the cameras, I can use software to make any file look like anything I wish... including matching of recent digital images with forty year old negatives. Today, if a Leica camera functions significantly better in some area (i.e. ergonomics, size, weight image) than it has value. If not, it is strictly pride of ownership.

Well, with my copy of Capture One Pro I can do a lot better than that -- but certainly not create the look consistently of another camera.

I think you are missing a few very real subtleties, n2, and subtlety is what works when we appreciate an image.

Direct link | Posted on Aug 31, 2015 at 20:46 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

Richard Franiec: Can anyone tell me why the Leica pictures showcased in the gallery are more special than what can be done with other manufacturers cameras/lenses?

I'm kind of having the thought that what is special is that the images are not 'special'.

They have a real feeling, rather than a 'camera' feeling, for someone who has lived in the areas photographed.

Something to think about?

Direct link | Posted on Aug 31, 2015 at 20:24 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

User9249539467: Up until the time that full frame digital cameras fell below the $3000 mark, there were many different reasons to choose a specific brand. My preference was Nikon because they were the only ones that paid attention to the color characteristics throughout their catalog; they all matched, and I shot mostly positive films. My goodness, Leica lenses were wonderful, but each lens was different. Now... we even have Sigma producing decent lenses! And the cameras, I can use software to make any file look like anything I wish... including matching of recent digital images with forty year old negatives. Today, if a Leica camera functions significantly better in some area (i.e. ergonomics, size, weight image) than it has value. If not, it is strictly pride of ownership.

Well, this is the dry-as-dust version of the must-put-down-Leica talk which always turns up.

Maybe you would like to take a look at the shooting experience notes also at DPReview, as they may answer your questions: http://www.dpreview.com/previews/leica-q-typ116/4

Also, I don't think those who use it are correct with the 'software makes any camera look I like' statement. If it did, would you want to manually work over every image, every lighting, every tonality, every subject?

Sometimes, and for those who like it, attention to detail and breadth in design are practical, as well as enjoyed, no?

Direct link | Posted on Aug 31, 2015 at 20:19 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)
In reply to:

rondom: OUCH...the raw files, when not loaded with a proper profile, show major corner vignetting at the edge...when cropped one gets the 28 mm, and the DNG records a slightly wider field of view, but wonder why?
and not only that, you already have two dust spots in your Q.
See this (please do not attack and answer constructively)
http://www.dpreview.com/galleries/5601265905/photos/3280945/untitled

This is the way better smaller cameras are designed today. The lens is able to be much more capable per size and weight etc. by not being asked to correct all distortion, vignetting, and color fringing. In-camera firmware does that, and very well.

Your raw developer will do so also, as soon as it's updated for this camera.

Direct link | Posted on Aug 31, 2015 at 18:31 UTC
On A lot to like: Real-world Leica Q gallery posted article (282 comments in total)

Finding this gallery particularly charming, before looking into the camera details.

But have you tried it on an iPad? It's looking fine on a laptop, but at least on my iPad 2 running iOS 9, things are seriously awry - you won't miss it...

Web guys doing things over the weekend, would seem, as there's been no problem with galleries until now...

Direct link | Posted on Aug 31, 2015 at 18:24 UTC as 51st comment
In reply to:

bovverwonder: Does the eye detection work when using the Metabones adaptor?

Interesting. You could sort of imagine Sony doing something micro-mechanical with the focus, but I think not really.

Maybe it is a matter of focal-length/field of view information being wrongly reported or looked up for the face firmware, so it's not looking consistently in the right place. Or as they say in German, oder?

In any case, sounds it may be a firmware upgrade that allows things to work, doesn't it?

Direct link | Posted on Aug 23, 2015 at 02:47 UTC
In reply to:

bovverwonder: Does the eye detection work when using the Metabones adaptor?

Did you actually try this?

Can't see it should make any difference at all, since face and eye detect work only with the resolved sensor image, and so should be fine.

Direct link | Posted on Aug 22, 2015 at 23:46 UTC
In reply to:

josikim: Does anyone think it's worth selling my 24mm f1.4 for this thing? I always had a hard time choosing between the 24 and 35, and this thing has both! I'm just curious if f2 will satisfy me and my love for primes. From the samples (Thanks Rishi), it appears to have a decent bokeh when shooting up close. And that sharpness, ouch (in a good way)... I love sigma's efforts in giving us common folks the best glass possible.

Btw Rishi, There's just too much shadow recovery going on in some of the pics. I'm sure the harsh sunlight didn't help, but it feels a bit unnatural. :)

Sounds good; had been considering that myself, learning the original intent. Take care.

Direct link | Posted on Aug 21, 2015 at 04:48 UTC
Total: 115, showing: 1 – 20
« First‹ Previous12345Next ›Last »