Friday, July 24, 2009

K-m Olive: Sharp JPEG, Soft RAW(PPL/SilkyPix)!

The pixel peeping continues! This time is for the in-camera JPEG against RAW with the K-m Olive.

First, here is the image directly out of the camera, in the finest quality JPEG (File at around 3MB), shot in the RAW+JPEG mode (PEF):-


(Downsized only, all EXIF data preserved)

And then, the directly ("automatically") converted image from RAW using the original Pentax Photo Laboratory (Version 3.60) software, saved as the highest quality (File at around 8MB):-


(Downsized only, all EXIF data preserved)

Now, the pixel peeping begins! Here is the 200% view crop from the in-camera JPEG:-


(Click to view in full: 100% crop, simple resizing to 200% (no re-sampling), saved as 100% Jpeg with IrfanView, for easy but accurate viewing here.)

And then the 200% view crop from the PPL (with SilkyPix image engine) RAW converted JPEG:-


(Click to view in full: 100% crop, simple resizing to 200% (no re-sampling), saved as 100% Jpeg with IrfanView.)

As it can be easily seen, the in-camera JPEG is obviously sharper, with higher resolution and contains more details. It wins hands down against the RAW converted image, which is also much larger in file size.

Just for more information, the original image was taken in Natural Custom Image mode in Saturation -1, Hue 0, Contrast +1, Fine Sharpness -1. The same settings have been automatically carried forward to the PPL by default. In order to confirm it is not a difference in the level of *sharpening* (generally called "Sharpness" for DSLRs' setting). I have tried to convert the same PEF RAW file using the maximum "Sharpness" setting, i.e., +4, to see if it could help. Here is the 200% view crop:-



Well, as it has been well expected, it doesn't help.

So, the SilkyPix RAW conversion engine of the latest PPL is really poor. It provides poor per pixel sharpness which the camera has already captured. It's just a waste of resolution and image details with such RAW conversion. I think other 3rd party RAW conversion software should be used if anyone is intended to shoot RAW with Pentax DSLRs. The bundled RAW conversion software just produced JUNK!

But then the problems with those 3rd party RAW converters are that:
1. They need extra bucks;
2. Many times and many of them are not as colour correct and the reproduced colours are not as faithful;
3. The compatibility is not as good!

It is sad and ridiculous as the bundled RAW software should produce better pictures, like what the Canon DPP can do, unfortunately, the PPL doesn't. But yet fortunately, the in-camera JPEG engine of the Olive K-m is very good. And actually, it is excellent, I would say. So, I just wonder, why Pentax did not use their own RAW conversion algorithms used in the Olive K-m to do the RAW conversion, even for the PPL?

More unfortunately, the same Soft JPEGs are yet found in the two K-7 production units I tested before. So, many times I just wonder what Pentax people are doing actually? I hope that they are not really like too much those Soft JPEGs, no matter in-camera or RAW converted (and the algorithms)!

Those Softy things are really rather unfortunate for all Pentax users and actually for Pentax themselves. Overall, it is just a very disappointing thing! :-(

Last but not least, the SilkyPix is infamous to produce low resolution and per pixel sharpness actually, which was what Klaus of the PhotoZone found about when he tested different RAW converters quite sometimes ago (now, I'm afraid that Klaus has deleted the review as I am unable to find it at the PZ):-

http://www.photozone.de/7Digital/rawindex.htm

He tested different Spatial Freqency Response of the RAW converted by different RAW converters using the Imatest.

But the evidence for its existence before:-
http://photo.net/digital-darkroom-forum/00Kwvi

Update (7-30) - Raw Therapee 2.4 Shootout Added:


(Click to view in full: 100% crop, simple resizing to 200% (no re-sampling), saved as 100% Jpeg with IrfanView.)

As seen in the above, the resolution and details are now obviously better than what the PPL 3.6 would deliver but still it is not as good as the in-camera JPEG owing to the more serious pixelisation resulted.

Also, the colour rendition and white balance are somewhat different than what in-camera JPEG and the PPL produces (more yellowish), as I have mentioned last time, the colour reproduction by third party RAW converters will not be as Faithful and would be less accurate.

Moreover, the RT Release Version 2.4 has compatibility issue or bug with K-m's RAW files. The program will crash each time when the RAW files are read in thumbnails. So, thanks to the hint by one of the above Blog pal who replied, I downloaded and tried the RC2 of the Version 2.4 instead and everything works fine. The interface of the RT is not so straightforward and direct for some of the key operations but the most serious problem of it I think is that it is very slow - both the DPP and PPL surely runs faster on the same PC machines that I use. It is an usable piece of software, though, and it is the only Free Standalone RAW converter that supports the Pentax DSLRs outthere (almost, AFAIK)!

In fact, the older versions of the Pentax Photo Lab (2.X etc.) were sharper, I just wonder why what is going on by now, actually?

Also, there is a little update regarding PhotoZone's RAW comparison links and report (which were changed), here they are:-

http://www.photozone.de/conclusion-on-going

http://www.photozone.de/Imaging

26 Comments:

Anonymous said...

Getting boring now. The Km is already as sharp at low isos as the Canon 5D, according to your tests.

You have run out of useful things to post since you have settled on the Km.

Pentax owners can save as Adobe Camera RAW if they want to, so what if you don't like Silkypix?

Olympus raw Converters show no improvement over shooting JPEG and Nikon Raw is poorly supported by in-the-box software in my D40.

In my experience, both Canon DPPP and Silkypix are good, and allow fine-tuning of such things as White Balance without worrying about it at the time of shooting. This is the main advantage of shooting RAW for many people.

Pixel-peeping is like looking at the Moon through a telescope: interesting, but outside the telescope we can't or don't want to get that close.

RiceHigh said...

I originally intended to see how RAW conversion with PPL could improve the Image Quality and Details.

But very unfortunately, I found the reverse.

The main point is shooting RAW should provide better IQ/PQ, but not the worse! Especially when the camera has captured more image data.

Also, just don't think that pixel-peeping is meaningless. The difference is actually visible on today's large and high resolution monitors and more importantly when the user is going to do cropping on original pictures!

But then, I have to ask again: WHY the RAW conversion with original software produce WORSE results? What's your "point"? What do you want to *argue* (with nothing)??

Anonymous said...

Your point is valid when viewing on large and high resolution monitors. No problem. Just like looking at the moon through a telescope.

It's just that you don't do that for hours. Eventually you either save the image on the storage device or print it. Unless we all print every picture at large sizes, there's no problem.

Michael said...

I have the K-m since Feb.09 now and run all my RAWs throug RawTherapee. Free, Easy to use, Crossplattform and produces high quality JPEGs. Since Version 2.4 it supports native PEFs like a charm! Try it, love it, use it!

Anonymous said...

Not sure I quite understand... are you telling that the K-m RAW output is no good or that the SilkyPix converter sucks?

As far as the "original software" issue goes, my understanding is that the most of RAW processors delivered by the camera manufacturers are delivered because they *need* to be delivered, but their quality ranges from solid to questionable.
Hence the market for third-party raw processors. You get what you pay for.

Why didn't you test on Adobe Camera RAW, for instance?

Michael said...

SilkyPix is trash, as we know. My Pana FZ-18 came with Silky, forget it. Some say its capable of a lot of things, but crappy interface and dead slow plus runs only under Windows. For us LinuxBoys no way.

BTW, what happend to the RAW vs. JPEG Debate? Any chance to get one going here :-)

Anonymous said...

Funny,

Just fired up Pentax Photo Laboratory powered by Silkypix on a Mac.

47n said...

APPLE APERTURE my friend !
no better software on MAC's !

RiceHigh said...

> Michael said...
> I have the K-m since Feb.09 now and run all my RAWs throug RawTherapee. Free, Easy to use, Crossplattform and produces high quality JPEGs. Since Version 2.4 it supports native PEFs like a charm!

I have just downloaded and installed the RT 2.4 on my Intel Dual Core Laptop but it crashed each time when I opened my above PEF file. Really dunno why.. :-(

Hate_Idiots said...

Hey, the Japanese love RICE.... Why don´t they eat up this guy ??????

Rice, my boy, I told you once, start writing about sex..
Don´t you like sex? Have you tried?

Well, just try, I love it !!!!!

Anonymous said...

RH,

w/ all due respect - but before claiming that this or that raw converter is bad you need to post the setting that were used in raw conversion - and see if people who are proficient in that specific raw converter tell you whether you in fact was using an optimal set of parameters... w/ SilkyPix it is actually the situation more than w/ other raw converters.

and I hope you were using the recent version of SilkyPix - SP v4 Release Candidate Pro ?

Anonymous said...

RH, the last version of RT v2.4 has a bug - you need to install the prev version 2.4RC4 (or RC3 - do not remember what was the last one before the final release).

Anonymous said...

Really? I thought the KM had higher resolution and was sharper than a Hasselblad CF Digital Back! I mean really. Sensor size does not apply to Pentax right? I mean why would the bigger, more mature sensors and software of the competitors produce better results?

Anonymous said...

Re: "Apple Aperture my friend."

Will check it out.

Thanks, Applepal.

Anonymous said...

Here's the revised missing links - looks like they redid the site:

http://www.photozone.de/Imaging

---------------

For me, the ultimate question is the as-recorded RAW; not the post-processed RAW via different software. Different makers quickly correct for noise and other factors, to add to the confusion. There are ample software choices for RAW processing; and you don't buy the software more than once, when you decide which you prefer.

-------------

For film, the first point of judgment was the negative, period. People could get different results when printing at the same size, based on englarger, lens, aperture, developing time, and technique. A "good" negative could be judged by the skilled eye, by looking through it.

Someone skilled could also deliberately "adjust" the negative processing by using different developers, temperatures, agitation, dilutions and times.

The pro strives to remove variables; or set up a consistent set of procedures if they run into difficult situations. The high-ISO capability of the DSLR offsets thise difficulties; as does a high frame rate allowing more "opportunities per second" - in a few cases 3x what the K20D can do. Ring-USM is what permits those high frame rates through very rapid focusing. Pentax improved the K-7 Imaging Engine - because video demanded it. I wish they'd paid more attention to the heat-related problems, as well.

Digital has similar "points" where changes can be made deliberately via changes of settings; with higher-end cameras allowing for more settings and range of variance. Some of those exceed anything that the ordinary shooter would ever need, just like the number of AF cross-points. Some people need them; and some don't care (but, perhaps they should ...).

Today, the question as to whether a camera model succeeds, or not, are about more than IQ. On the high-end, MP count is still a race; as is FPS capability. Photosite size is a dominant IQ factor, affecting noise levels as well. Today's lenses 3 stops down can likely produce a good enough image for nearly anyone, and the pro-level zooms are right up there with the primes - with the primes ahead at large apertures, and edges.

Amvona has a very nice aluminum alloy 1-pound monopod for $19.99 that can support 8.8 pounds, and that will help a lot with stabilization even at long shutter speeds. The photographer has to ultimately be responsible - a novel concept, apparently.

http://www.amvona.com/

Dufus arse said...

Why not just add 1)silkypics version:
3.03
AND a quote:
Conversion Results
A verdict for the conversion quality with SilkyPix isn't easy. You have to work quite a while with the application to explore all the possible tuning options.
Out-of-the-box SilkyPix does A LOT of image manipulation resulting some of the softest default outputs within the scope of this overview. However, it is quite easily possible to persuade the application to perform a much less aggressive approach. For low ISO images it is certainly a good idea to disable the luminance noise reduction which is active by default. You can choose between two different sharpening algorithms - "normal sharp" and "pure detail". It is difficult to judge the effects between them because their sharpening characteristic is quite different at identical settings - "normal sharp" produces a lot more edge sharpening for instance. Apart from these (supposedly more sophisticated) sharpening controls you can also add an additional unsharp masking (USM) on top at the time you export the tuned images to disk. This is a quite strange approach because if taking advantage of USM you have to review the resulting effect on your images twice. "De-mosaic sharp" is another control option but I failed to see a really significant visual effect regardless of the setting. All-in-all it is possible to achieve a pretty sharp output without reaching ACR or RawShooter though.

A. Rohman said...

Nice blog. stop dreaming start action

Le, Dave D. said...

Help, I have the regular black K-m/k2000. I was wondering if there are any improvements on the sensor of the K-m Olive? Or are the colors just different?

Anonymous said...

The Olive edition is limited release overseas, and should be identical to the K2000 in specs. The white edition also had some special finishes on lenses to go with it.

http://www.ok1000pentax.com/2009/03/camo-green-km-available-now-in-japan.html

http://ricehigh.blogspot.com/2008/12/star-wars-stormtrooper-k2000w-le.html

K20D also had a Titanium edition.

Panasonic Lumix G-1 was marketed in 3 body colors - red, blue, or black.

Anonymous said...

More collectibles and specials:

http://www.slashgear.com/pentax-ap50-dslr-special-50th-anniversary-edition-015535/

http://www.digicamreview.com/2008/09/pentax-digital-slr-covered-in-bling.html

http://www.luxurylaunches.com/other_stuff/asahi_pentax_lx_gold_worlds_first_waterproof_and_dustresistant_slr_camera.php

http://gizmodo.com/216979/correction-pentax-solid-gold-slr-is-actually-from-1981

One company had custom leathers for the analog SLR's:

http://www.cameraleather.com/pentax/

RiceHigh said...

> Le, Dave D. said...
> Help, I have the regular black K-m/k2000. I was wondering if there are any improvements on the sensor of the K-m Olive? Or are the colors just different?

My impression is that the K-m Olive has a better JPEG engine, this has no scientific proof or test, though. As for colours, they are quite similar.

Anonymous said...

Pentax has released a new version of the PPL-software, with improved performance. It is not yet available for download as an upgrade, but it will be soon. Currently, it ships with the K-7 but will be available later for other Pentax DSLR's.

The new version of PPL has much improved interface too and enhanced features and functions.

Michael said...

Hm, could you post next time the sharpening parameters you used for RawTherapee? I usually sharpen with smaller radius (0.6-0.8 px) and halo suppression and need some tries to get optimum results, but after that I usually get sharper results with more resolution then the In-Camera JPEGs.

Michael said...

... oh btw. before I forget:
In RT you can set different Demosaicing-Algorythms wich sometimes help to avoid demosaicing errors and push some critical resolution....

RiceHigh said...

I have not changed any setting in the RT. I converted the picture using all default (or automatically calculated) values.

Anonymous said...

Blue here. You don't have to use SilkyPix. I use mostly PEF and CS3.

Post a Comment

Related Posts

 
Creative Commons License
RiceHigh's Pentax Blog by RiceHigh is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.