Jump to content
Sign in to follow this  
sergiolov

JPG resolution loss

Recommended Posts

Guest stnami

Advertisement (gone after registration)

Sol is pretty much on the money here it's like buying a sports car and using a performance Inhibitor........... despite that a camera's processing engine is not going to treat every image taken on its merits the first instance

Share this post


Link to post
Share on other sites

Rob A--

Just a suggestion: Instead of posting four consecutive entries on the same topic, why not just append the comments to the original? Your posts 49-52 above all came within 16 minutes, well within the hour allowed for editing or deleting the original message.

 

 

All--

Already, the M8 writes DNG images faster than it writes JPGs. Might asking for less-compromised JPGs slow down the JPG processing? Then we'd have a new bone to pick.

 

Don't know, only speculation.

 

--HC

Share this post


Link to post
Share on other sites
Rob A--

 

All--

Already, the M8 writes DNG images faster than it writes JPGs. Might asking for less-compromised JPGs slow down the JPG processing? Then we'd have a new bone to pick.

 

Don't know, only speculation.

 

--HC

 

On the other hand, perhaps with less compression it would be quicker?

Share this post


Link to post
Share on other sites

Sergio - Interesting that the D2x/M8 side by side shots are, it is not an apples to apples comparison. The D2x files are good, as I would expect, but both D2x show files show sharpening halation whilst the M8 files show none. I shoot Dng + Jpeg and use the Jpeg for fast editing, it could of course be a lot better but I agree with the point that Howard alluded to; I would rather the Jpeg stay poor than it be improved at the cost of increased file writing speed.

 

....................Chris

Share this post


Link to post
Share on other sites
Sergio - Interesting that the D2x/M8 side by side shots are, it is not an apples to apples comparison. The D2x files are good, as I would expect, but both D2x show files show sharpening halation whilst the M8 files show none. I shoot Dng + Jpeg and use the Jpeg for fast editing, it could of course be a lot better but I agree with the point that Howard alluded to; I would rather the Jpeg stay poor than it be improved at the cost of increased file writing speed.

 

....................Chris

 

Hi Chris,

no sharpening was applied to the images. What you see on the vertical lines is moire + aliasing due to the lens outresolving the sensor. I don't know how to perform a more apples to apples comparison than this. As Jono said, I think that a less compressed JPEG should write faster. But at the end of all, if Nikon and others are able to obtain reasonably good jpgs,(resolutionwise), why should'nt Leica at least try?

Sergio

 

(a second thought- what if Leica reduced the jpeg resolution quality on pourpose, to mask moire and aliasing due to the lack of the AA filter?)

Share this post


Link to post
Share on other sites
...(a second thought- what if Leica reduced the jpeg resolution quality on pourpose, to mask moire and aliasing due to the lack of the AA filter?)

Good point, Sergio--

LFI and this forum have spent a lot of time dissecting the M8's version of DNG, but almost none trying to decipher Leica's choices in their JPG engine. This thread may be a good start in that direction.

 

--HC

Share this post


Link to post
Share on other sites

Advertisement (gone after registration)

Sol is pretty much on the money here it's like buying a sports car and using a performance Inhibitor........... despite that a camera's processing engine is not going to treat every image taken on its merits the first instance

 

The post isn't about why you should want to shoot jpg, it's about why the JPG files are too compressed if you DO want to shoot jpg.

Share this post


Link to post
Share on other sites
On the other hand, perhaps with less compression it would be quicker?

 

Jono--

I thought about that and ran a quickie and very interesting though possibly inconclusive test:

 

Test setup:

Transcend 4GB 150x SD card, formatted in camera, 234 images total on card after test; 955.4 MB free on card after test.

M8 in auto exposure mode, ISO 2500, -2/3 stop exposure correction, AWB. Battery showing full charge after test.

Tri-Elmar at 50mm, f/4. M8-determined exposure 1/180 sec.

Camera on tripod, facing rather evenly illuminated off-white surface at about 1 m, lens focused to infinity.

 

Test procedure:

Set camera to continuous exposure mode; hold down release button for ten exposures.

Time from first pressure (first exposure) until the red 'busy' light on rear of camera quits blinking. (Used wristwatch, not stopwatch; timings approximate.)

 

At DNG setting, procedure took 45 sec. File size 10.08 MB. Total 100.9 MB on card.

At JPG Basic, procedure took 48 sec. File sizes 4.67, 4.68, 4.69, 4.70 MB. Total 46.9 MB on card.

At JPG Fine, procedure took 50 sec. File sizes 4.66, 4.67, 4.68, 4.70, 4.71 MB. Total 46.8 MB on card.

 

Interesting that JPG Fine and JPG Basic files are all in the same size range. I assume that is because there were basically no details to render.

 

Even though JPG Fine files take up slightly less space on the card, it took slightly more time to generate and write them than the JPG Basic files.

 

Others might want to do similar tests, such as:

testing cards of various brand, size, speed;

testing cards in empty, half-full, nearly-full status;

testing thru buffer-full exposure cut-off and continuing thru buffer-clear indication;

testing a subject of some detail;

testing with various battery states, ISOs, white balance settings;

etc.

 

--HC

Share this post


Link to post
Share on other sites
....no sharpening was applied to the images. What you see on the vertical lines is moire + aliasing due to the lens outresolving the sensor....

 

Sergio - I did not doubt for one moment that you had not applied sharpening. Where I see [undisclosed by Nikon] in-built unsharp masking in the D2x files, you see moire + aliasing. I am genuinely interested to learn what you are seeing that I am not.

 

...............Chris

Share this post


Link to post
Share on other sites

Does anyone understand jpg compression? Does more compression mean that the algorithm repeats through cycles to convert adjacent pixels, or does the algorithm just adopt a larger ingterger into a calculation? Where are the rate determining steps in writing a jpg to card?

Share this post


Link to post
Share on other sites
Does anyone understand jpg compression? Does more compression mean that the algorithm repeats through cycles to convert adjacent pixels, or does the algorithm just adopt a larger ingterger into a calculation? Where are the rate determining steps in writing a jpg to card?

 

VERY broadly, what the most common form of JPEG compression does is to take 8x8 blocks of pixels, and build what is effectively a mathematical model of each block. The nature of the mathematical model (which is a discrete cosine transform, or DCT) is that what it delivers is also an 8 by 8 set of numbers, but those values have the property that the most useful information is in in one of those numbers, and less and less information in the remaining numbers. So you can get a very low quality image by dropping every number except the first, or an effectively identical image with all of the numbers. What the quality setting on a JPEG does is to decide how many numbers you use.

 

Depending on hardware, the rate limiting step is usually the DCT.

 

Sandy

Share this post


Link to post
Share on other sites
Jono--

I thought about that and ran a quickie and very interesting though possibly inconclusive test:

 

Test procedure:

Set camera to continuous exposure mode; hold down release button for ten exposures.

Time from first pressure (first exposure) until the red 'busy' light on rear of camera quits blinking. (Used wristwatch, not stopwatch; timings approximate.)

 

At DNG setting, procedure took 45 sec. File size 10.08 MB. Total 100.9 MB on card.

At JPG Basic, procedure took 48 sec. File sizes 4.67, 4.68, 4.69, 4.70 MB. Total 46.9 MB on card.

At JPG Fine, procedure took 50 sec. File sizes 4.66, 4.67, 4.68, 4.70, 4.71 MB. Total 46.8 MB on card.

 

Interesting that JPG Fine and JPG Basic files are all in the same size range. I assume that is because there were basically no details to render.

 

 

--HC

Hi Howard,

Those are interesting file size numbers. 1/100MB difference for a low detail subject looks like an oops, but it is probably just the compression configuration.

Bob

Share this post


Link to post
Share on other sites

Bob, I agree. The results are quite strange and may be the result of an error.

 

Someone else ought to re-try it; I'm guessing that it's because there was no detail (out of focus image of a white wall) to record; thus when the JPG algorithm checks for values, they're all the same and can be nicely compacted.

 

But the test needs verification (similar circumstances) and it needs to be checked against images with detail (particularly since that's what the thread is about).

 

Nonetheless, "oops" or no, it took longer to write the JPG Fine files, which actually took up marginally less space on the card. Possibly a function of where on the card the files went (would top of empty card write more quickly?), possibly a function of the card programming (4 GB SD, not SDHC). Who knows?

 

My very tentative conclusion is that the JPG Fine files took longer because it took longer to process them. If that's the case, it would likely take longer yet to generate a more detailed JPG Finer file. But the difference is minimal, so we might never notice.

 

Certainly needs more research.

 

--HC

Share this post


Link to post
Share on other sites
Sergio - I did not doubt for one moment that you had not applied sharpening. Where I see [undisclosed by Nikon] in-built unsharp masking in the D2x files, you see moire + aliasing. I am genuinely interested to learn what you are seeing that I am not.

 

...............Chris

 

Chris, I was referring to this artefact. I have difficulty in identifying sharpening artifacts in the d2x images. More noise, yes: 1 1/2 stop more.

Share this post


Link to post
Share on other sites

Sergio - Thanks.I see a small amount of moire, but mostly I see a white edge to the black bars; halation [as in halos presumably]. I only know of halation in connection to un-sharp masking, which is why I wondered if the D2x files had some undisclosed sharpening in them.

 

.................Chris

Share this post


Link to post
Share on other sites

Sandy, thanks for the explanation, sounds interesting. The DCT sounds like the Fourier transforms I vaguely recall studying at school

Where can I read more about JPG compression, image algorithms in general, and DCTs?

Share this post


Link to post
Share on other sites
Sandy, thanks for the explanation, sounds interesting. The DCT sounds like the Fourier transforms I vaguely recall studying at school Where can I read more about JPG compression, image algorithms in general, and DCTs?

 

Carsten,

 

Actually, that's exactly right - the DCT is a transform into the frequency domain, and what the various quality levels do is to drop more and more high frequency information.

 

More info:

 

Photo.net - Jpeg Compression by Gordon Richardson - photo.net

JPEG Compression

http://www.fho-emden.de/~hoffmann/jpeg131200.pdf

 

Sandy

Share this post


Link to post
Share on other sites
Bob, I agree. The results are quite strange and may be the result of an error.

 

Someone else ought to re-try it; I'm guessing that it's because there was no detail (out of focus image of a white wall) to record; thus when the JPG algorithm checks for values, they're all the same and can be nicely compacted.

 

But the test needs verification (similar circumstances) and it needs to be checked against images with detail (particularly since that's what the thread is about).

 

Nonetheless, "oops" or no, it took longer to write the JPG Fine files, which actually took up marginally less space on the card. Possibly a function of where on the card the files went (would top of empty card write more quickly?), possibly a function of the card programming (4 GB SD, not SDHC). Who knows?

 

My very tentative conclusion is that the JPG Fine files took longer because it took longer to process them. If that's the case, it would likely take longer yet to generate a more detailed JPG Finer file. But the difference is minimal, so we might never notice.

 

Certainly needs more research.

 

--HC

Hi Howard,

We actually had a moment of sunshine today and I ran a few shots to compare Jon's lower saturation idea. I found that it did work with my tone masking and bringing the saturation up in my editor was better, because it preserves the highlight saturation. It could be that there is some color "near clipping' going on at normal saturation.

If we get some breaks in the weather, I'll check out the basic/fine on something with detail. The write times don't bother me for my usual shooting pace.

Bob

Share this post


Link to post
Share on other sites
Sergio,

Good comparison, but you have to set the D2X to ISO 200 for a proper comparison as ISO 160 on the M8 is the equivalent of ISO 200. That explains why the D2X shot is brighter in this comparison.

Didn't think the D2X would look sharper in RAW but it does and a LOT better in JPEG.

 

Isn't it the opposite? The ISO of the M8 needs to be raised to lighten its image to match the exposure from the D2X. So ISO 160 on the Nikon may require ISO 200 or so on the M8. (But an M8 only has full stops so you'd have to lower the ISO on the Nikon.)

 

I'm curious what kind of detail the Nikon would have at ISO 100.

 

While this test seems pretty tight for what you were trying to discover (raw vs. jpeg), maybe a test of both cameras at their lowest ISOs using f8 and varying the shutter speed until the histograms match would allow for maximum detail comparisons. (I'm not sure why the M8 image has so much less detail in the cable in the foreground - longer focal length, focused further back, lack of resolution?)

 

Plus I think one would need to set the Nikon lens a bit wider to match magnification.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
Sign in to follow this  

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue., Read more about our Privacy Policy