Trials

[nicolas claris]

The season for boats photo shoots will arrive soon, so I need to get back to the MF way of shooting (one calls that habits) I left at the end of 2088 when I sent back the Sinar ti the factory.
Rereading what I and others wrote in the "Shooting with a Sinar" fora did help me a lot…
However, the Pentax 645D is a completely different machine than the HY6.
One doesn't hold it the same way, there's no back, the view finder is much smaller but at least right is on the right and not to the left! The menus are more dlsr like (a good point!) and the rear screen is pretty accurate!
However 2 one must be careful, IT IS NOT a DSLR! It looks like one, you almost feel like having one in hands but it needs the same care as all MF cameras. Light and speed are needed.
Thread after thread I'll surely let you feel the differences and my findings, can't write a full description in one place at one time… sorry
Then, the next thing is to start training in order to feel having the camera embedded in myself, forgetting the need to remember where are the different buttons, settings etc… being able to automatically set the camera upon to my request for what I want to achieve… This takes time and while it is always possible to do a "side job" when shooting for the clients, this is not very productive and my clients do not pay me to spend paid time for my learning.
Therefore I need to use the rare free time I get to jump on the Pentax 645D (and believe me this is not a punishment!).
Last week-end we took some days off with Marine and went to Rome (Italy).
This city is magnificent!
I knew that so I took the Pentax 645D with one lens only, a SMC D-FA 645 55MM F/2.8.
This lens is delivered as a combo with the 645D and believe me, it's a pure jewel!

Walking in the streets, we got into a market via Cola di Rienzo and I found that it would be interesting to see if the 645D would be able to deliver all these colors!

Hand held: ISO 400 - ƒ 6.3 - 1/50 - Aperture priority

100% crop:

Of course yes!

Full size image

 

[nicolas claris]

After many trials back and forth between LR4 beta and C1 6.3, I have decided to give a serious go to LR4. It renders much better the smooth gradients and blurred (OOF) parts.
I certainly need to practice more C1 wit the Pentax 645D files, but for now I'm faster with LR4 (which by the way takes some times to digest the raw file (wether they are PEF or DNG) to deliver the 240 Mb of the 16 bit tif files…
I also take into consideration that LR brings all the lens correction needed as it has the Pentax 645D lens profiles embedded …

So today was time to test and try the smc FA 645 80 - 160 mm f/4,5
But before I ran all tests to check the focus precision (The Pentax 645D has the possibility to set between -10 to +10 front/back focus) but for both 55mm and 80-160 the best setting was to zero.

One of main need for this lens is to catch details of wood/leather/aluminum/steel work by yacht fine craftsmen.

So I shot… my bookshelf!

first at 80 mm (64 mm 24x36 equivalent)
Tripod: ISO 200 - ƒ 16 - 0.4s - Aperture priority

100% crop

Full size image


Then at 160 mm (128 mm 24x36 equivalent)
Tripod: ISO 200 - ƒ 16 - 0.3s - Aperture priority

100% crop

Full size image


​

 

[Cem_Usakligil]

Hi Nicolas,

This is a very interesting topic and I am looking forward to reading more about the 645D as you progress further with the camera.

The texture in the examples you have shown is quite good so the camera will serve you certainly well in capturing the details of wood/leather/aluminum/steel, etc. The rendition of colors also seem great. Of course, since the sensor does not have an AA filter, there is some color moire to be seen (in the smallest details) and also some stair-stepping in lines. But there are many PP methods which can be used to deal with these issues as you know.

I have had the opportunity to hold the 645D in my hands and have shot a few frames. It may look like a large DSLR but it is indeed a different beast. I think that it is a good idea of yours to try and get used to the camera as much as possible before using it in the field.

Please show and tell more when you have the time!

 

[Asher Kelman]

Nicolas,

I'm so pleased that you're testing the Pentax 645D. This camera appears to be a breakthrough in price and capability. I'm hoping that your work reveals it's qualities as I know that many of us are interested in this new offering.

I have questions on AF. Is there live view? Can one focus using outlines of edges illuminating when in focus?

Also in the focus corrections for zoom, can one use more than one adjustment at the extremes of the zoom?

Asher

 

[nicolas claris]

Hi Cem
Thanks for the comments.
Obviously there is still a lot more to learn about PP and get even better results.
Today I have made some good progress…: ) particularly with 5 toned maps shots.
Also LR is not my usual software, it still delivers better results than C1 (which also likes to crash when working with Pentax DNGs!).
I have also discovered that in camera DNG -when used with C1- is less accurate and full of CA with wide angles- than PEF file converted to DNG with Adobe DNG converter… go figure!
The suggested software by Pentax is a special version of Sylkipix, I tried it but it doesn't seem as good as LR or C1 and is very slow. Also I don't wish to add a 3rd PP software to learn!

Hugs to Holland (and Turkey!)

 

[nicolas claris]

Bonsoir Asher
The main difference - and that's a huge one! - is that on the contrary to Sinar, I bought the Pentax!
So it's a test but needed before going to work! I have to convince my accountant that the 645D is not only a toy ; )

No, unfortunately there is no live view.
There are many focusing screen available but the standard one is… standard.

The AF is quite fast (and silencious) and has 11 points.

For interior shots, I use manual focusing with focus confirm and it works very well.

There is only one adjustment per lens, prime or zoom.

Hugs to L.A.!

 

[Bart_van_der_Wolf]

Hi Nicolas,

Exciting stuff, and I'm looking forward to your findings as well.

No, unfortunately there is no live view.
There are many focusing screen available but the standard one is… standard.

Is there no Live View, or just no tethering? The tethering will apparently be made possible, unfortunately at a cost of US$ 200, in March 2012:
http://www.dpreview.com/news/2012/02/07/Pentax_645D_transmitter_tethering_software

I couldn't find the press release on the Pentax website, hence the link to DPreview.

Cheers,
Bart

P.S. some 'translated' additional info from the Japanese website

 

[Bart_van_der_Wolf]

I also came across a planning overview of new lenses:

Cheers,
Bart

 

[Asher Kelman]

I also came across a planning overview of new lenses...

Bart and Nicolas,

I wonder whether Ricoh management will aggressively revise that schedule for the lenses? Do they need to wait? What lenses are really needed for this system to seduce pros and enthusiasts?

Asher

 

[nicolas claris]

Hi Bart and Asher

here you'll find the actual range of Pentax lens:
http://www.pentax.fr/fr/optiques-645.html

I do have a loan copy of the 25mm but it is a test lens only, the real one is afaik not on the store shelves for now…

AFAIK only this 25 mm and the smc D-FA 645 55mm f/2.8 (which I also have) are new lens dedicated to digital.
I also have a smc FA 645 80 - 160 mm f/4,5. It is not digital but perform quite well, see my bookshelf above!

So one has to be careful buying a non digital lens for this camera, I guess a simple test in the brick and mortar shop should be enough if it is worth waiting or buying.

Personnaly, for my kind of work, I don't think I'll jump up to more than 200 mm unless it is stabilized…

Regarding the live view, no definitively not on this camera, I wonder if it could be enable later via a firmware update…?
Tethering is not yet arrived but I don't miss really this feature as it is not in my workflow…

 

[Michael Nagel]

Hi Nicolas,

playing around is one of the most efficient ways to learn, so don't worry and carry on. Playing around does not mean that it must be a toy you are handling. I am curious on the next results.

It is a shame that the 25mm did not arrive yet and I hope this will soon be the case.

Personally I do not take the 'digital' designation for a lens too seriously. Often this refers to the coating of the rear lens, which has become common practice before the digital era. The best is to try anyway, but I would not worry too much for the newer pre-'digital' lenses.

Happy playing then!

Best regards,
Michael

 

[nicolas claris]

Bonsoir Michael
Be assured and don't worry ! I assume the fact that I bought this camera end enjoy the learning curve!
I'll use it next Sunday on an important shoot for a new client… hmmmm I'm anxious and delighted in the same time!
I have to admit I have been very positively surprised by the quality of the 80-200 which is not digital dedicated…
I'm a happy player!

KInd regards

 

[Jerome Marot]

I don't own a medium format camera, but I use older, non-digital lenses on my DSLR. I never noticed any adverse effects. Reportedly, the only difference is the coating of the back lens which, on pre-digital versions, could reflect the shiny sensor. The pictures would then have a greenish hot spot in some circumstances. I suppose that said circumstances are rare, because I have never noticed it.

This being said, optics have improved in the past 20 years and some older designs suffer from defects which are rather obvious with the higher resolution of digital sensors and with the fact that we tend to look at pictures with much larger enlargements than we used to.

 

[Asher Kelman]

Nicolas,

What is the max size at 300 dpi uninterpolated and then what would you use as software today to enlarge the image further if you needed to. I do know that you have been very successful in the past printing 3 meter hight pictures using the 1DSII and probably the 1Ds III too. Here the Pentax likely has better pixels. I'm hoping this will be obvious in your giant prints.

Any ideas on this?

Asher

 

[nicolas claris]

Hi Asher

giant prints are one of the reasons I chose the MF route.

Size of files is: 7264 x 5440 pixels
that is @ 300 dpi: 61,5 x 46,06 or 24,213 x 18,133 inches

As for enlarging, I've used different plugins or processes in the past, but will follow Bart's suggestion to use photozoom as the tests I've done are a bit better…

PS the 1Ds Mk III is still in my bag!

 

[nicolas claris]

This being said, optics have improved in the past 20 years and some older designs suffer from defects which are rather obvious with the higher resolution of digital sensors and with the fact that we tend to look at pictures with much larger enlargements than we used to.

Hi Jerome
you have pointed in the right direction!
The important thing regarding lens is not the fact that they are mad for digital but that they use the best latest technologies.
Lens are so important! they are the door where the light enter first!

 

[Bart_van_der_Wolf]

As for enlarging, I've used different plugins or processes in the past, but will follow Bart's suggestion to use photozoom as the tests I've done are a bit better…

That's right. I've recently done many detailed comparisons, and for subjects with lines and high contrast edges, Photozoom Pro (which I've mentioned before in my posts) does something remarkable, it creates additional resolution! Edge transitions can be sharper than the number of pixels added in the interpolation process, and there are no halos.

How it performs with images from a sensor without AA-filter remains to be seen, but it does have some controls that will reduce jaggies a bit. There is of course also an option to pre-filter the image, take away a smidgen of the hyper-sharpness that will cause issues, and produce a better upsampling result.

I suppose later in this thread, when Nicolas e.g. has some images showing ship rigging against the sky, we could add a 645D crop of an example resized to wall size.

Cheers,
Bart

 

[nicolas claris]

I suppose later in this thread, when Nicolas e.g. has some images showing ship rigging against the sky, we could add a 645D crop of an example resized to wall size.

Here we are!
No need to go at sea for this, but just needs to lean out of my office window and set the 80/160 mm on the 645D.

Pentax 645D - smc PENTAX-FA645 80-160mm F4.5
160 mm - 1/500 s - f/10 - ISO 200
​

@ Bart: The corresponding DNG file is on its way for you to show enlargement possibilities on this 645D file…

 

[nicolas claris]

100% crop

100% crop (the thin red line at the back of the mast is not any defect, it is a thin red rope (will call it a messenger (will be used later to pull the "real" stronger/heavier rope)

100% crop

100% crop

100% crop

 

[nicolas claris]

Downsampling

Hello all
a few days ago, Bart has posted very good informations (there).
If for ages upsampling has been crucial for many of us wishing to get giant prints, I'm now also facing with downsampling (mainly for the web and low resolution slideshow for client's selection)!
Reducing with good quality a 7264 x 5440 pixels file to 800 x 599 pixels seems a real challenge…
Any hints?

 

[Bart_van_der_Wolf]

If for ages upsampling has been crucial for many of us wishing to get giant prints, I'm now also facing with downsampling (mainly for the web and low resolution slideshow for client's selection)!
Reducing with good quality a 7264 x 5440 pixels file to 800 x 599 pixels seems a real challenge…
Any hints?

Hi Nicolas,

I'm not too familiar with software for the Mac platform, but I believe you also have a Windows machine used by one of your employees for Web related stuff. In that case there is a nice (free) application that does a nice job, called RIOT. One of the useful features is that image parameters can be tweaked for quality (e.g. the JPEG subsampling of color in addition to the overall compression/quality setting), but also for size in bytes (for webpages that load faster or consume less bandwidth on mobile devices. It offers several downsampling filters that can reduce some of the issues that accompany downsampling, e.g. moiré and jaggies.

Unfortunately, it is not perfect. For one, it is missing one parameter that improves images in a sometimes subtle (and sometimes in a dramatic) way, gamma control. When the resampling is not executed at gamma 1.0 then there may be issues such as described at:
http://www.4p8.com/eric.brasseur/gamma.html . One retains more sparkling highlights when the gamma is reduced to linear gamma for the downsampling interpolation itself.

The only application (a set of commandline utilities) I know of that allows full control, is Imagemagick. One can make a script to easily execute the right steps for a given task, and e.g. in Windows create a batch file where images can be dragged and dropped on to produce the required output.

Cheers,
Bart

 

[Cem_Usakligil]

I agree with the comments of Bart and would like to add that the downsizing by LR is very good as well. Did you try it yet? If you want to stick with PS, you could apply a Gaussian blur first (the diameter of which being related to the factor of downsizing required, Bart knows the formula). Then downsize using bicubic and sharpen again for output.

 

[nicolas claris]

Thanks Bart and Cem
unfortunately RIOT is only for Windows (as mentioned) and the Windows computer we have is running all days long!
Imagemagick has a Mac version (I have to understand how to install it…; )
But as suggested by Cem, Ill give try to LR.

Kind regards

 

[Asher Kelman]

Bart, Nicolas and Cem,

This discussion is so valuable and important. We need a master reference section to all these good ideas. I keep forgetting all these gems.

Asher

 

[Bart_van_der_Wolf]

Pentax 645D Read Noise characteristics

Hi folks,

I've been looking at some 'Black frames' from a 645D, in an attempt to make an estimate of the shadow noise performance (which is typically dominated by 'read noise') and as a consequence of the dynamic range of the Kodak KAF-40000 sensor that's supposedly used in the 645D.

I used a new software tool for most of the analysis of the Raw file data, Rawdigger (Beta), before demosaicing.

A common method to determine the read noise, is by using so-called black frames. A black frame can be produced by 'exposing' a frame with no light, body cap (or lens cap) in place, and the viewfinder covered. One normally also uses the shortest possible shutterspeed (1/4000 sec. in this case) to avoid temporal noise from building up during the exposure. Thus we get a Raw data frame that will practically only consist of the noise from the readout cycle, reading the analog signal from the sensels and quantization by the Analog to Digital Converter (ADC).

The most accurate data can be produced by subtracting 2 frames from each other because that will eliminate any pattern noise influence, and will leave just random noise. However, the 645D stores the data with an offset that clips approx. half of the noise as shown in the graph below (the X-axis are the Data Numbers (DN) as recorded in the file):

As one can see, the Read Noise doesn't show a Gaussian bell curve but roughly half of it.

Therefore normal evaluation of the standard deviation based on the above info is not possible, and we cannot be certain that the peak (which I clipped in the Y-axis) at 0 represents exactly the missing (negative) DN bins (+/- the 0-bin), or that the distribution is symmetrical. The alternative method to get the data is by taking many underexposed bracketed images, and extrapolate the values we find there by fitting a regression curve. I didn't really have the time to evaluate that, and to have someone produce all those files.

Luckily, by inspecting the non-image pixel data surrounding the actual image data in the Raw files I noticed 2 areas with 3 rows of pixels that seemed to be not photographic exposure, but system generated data (thus without photon shot noise). The readout process did add some noise which looked exactly like the visible part of the read noise shown above. In fact it also changed exaclty the same with increaded ISO settings, so I decided to use that data of which I summarized the resulting noise here:

Readnoise estimate

ISO	R	G	B	G2	approx.Avg.	
 100	6.1	6.2	6.0	6.2	6.13
 200	10.1	10.0	10.1	10.2	10.1
 400	18.8	19.0	19.7	20.0	19.4
 800	36.8	37.3	36.9	38.3	37.3
1600	77.6	79.1	76.8	79.7	78.3

These values are for 14-bit Data Numbers, so for comparison to other numbers one may need to e.g. multiply all numbers by 4 for 16-bit values.

An interesting observation is the small systematic higher response for the G2 channel compared to the G1 channel. This means that for the best quality Raw conversion there must be an internal calibration between these channels to avoid demosaicing noise structures. The systematic differences between both green channels and the Red and Blue channels can be addressed together with the color balancing done during the Raw conversion. It just demonstrates that a good Raw converter will get better results from these files.

ISO 100 and ISO 1600 are special extended ISO settings, but as one can see, it does pay to use ISO 100 if one wants to reduce Read Noise. Since the saturation value at all ISO settings of the 645D remains at approx. Digital Number (DN) 16064 from the 14-bit ADC, the Dynamic range is maximized at ISO 100.

To determine the actual Dynamic Range (engineering definition) at a given ISO, one divides the signal level at which the sensor saturates (16064) by the standard deviation of the Read Noise and takes the Logarithm base 2 of that ratio to express the DR in stops.

This gives the following Dynamic range numbers:

ISO  100 = 11.4 stops
ISO  200 = 10.6 stops
ISO  400 =  9.7 stops
ISO  800 =  8.8 stops
ISO 1600 =  7.7 stops

As can be seen, roughly 1 stop of DR is lost for each full ISO step.

That's it for the moment, but there is more info to share, so stay tuned ...

Cheers,
Bart

P.S. According to the Kodak Datasheet for the KAF-40000 sensor, the estimated linear dynamic range is 69.3 dB, which equates to 11.51 stops.

 

[nicolas claris]

That's it for the moment, but there is more info to share, so stay tuned ...

Wow! thanks Bart!

I do stay tuned ;- )