What's new
Medium Format Forum

Register a free account now!

If you are registered, you get access to the members only section, can participate in the buy & sell second hand forum and last but not least you can reserve your preferred username before someone else takes it.

H3D

He Austin!

QUOTE biggest. It depends on when one is talking about, and by what metric "biggest" is measured. I've worked as an engineer/scientist for IBM, DEC, Compaq, HP and Sun. ENDQUOTE

For me you should substitute Philips Information Systems, DEC, Compaq and now HP. No IBM and Sun sofar ;-)

Funny.. this.. Wilko
 
Gentlemen,

Can we please refrain from interpersonal bickering on this list? Take it into personal email if you must.

And Q, what you are telling Austin not to do is *exactly* what you do constantly. Pot kettle black?

Anyway, first and last comment from my side on this one. I do not want to contribute to another thread from hell.

Wilko
 
"And many self proclaimed experts as well. "

LOL.

Austin, you love to pour over how the sausage is made ... while others just taste the sausage and decide if it's good or not.

Specsmanship is also a hoot. As if a photographer is going to drop 40 grand based on marketing specs like a consumer buying a digital point and shoot. Hasselblad could have said their backs are 32 bit and I'd take their word for it ... until it's time to buy ... then the proof is in the pudding whatever recipe they may have used.

I'm not an engineer. I don't want to become one either. I am grateful there are engineers to design the gear I need. If some engineer says one thing and another says another, who cares? DOES the camera deliver what they imply with all the techno babble? That's for the photographer to determine, not an engineer... and no amount of test charts, double talk or manufacturer spin is going to alter that fact. If a manufacturer touts this and that, and the actual work doesn't live up to the techno promise they will pay the piper eventually.

Personally, while it is fun on these forums, I really don't need to know how the sausage is made, or want engineer anything ... I make photographs.
 
Simon:

Erwin Putts also wrote the following about the M8. "My view is this: the image quality that the M8 delivers and the sheer joy of using the M camera, is the best argument to select the M8 as the main camera. (I sold my 5D). And the M7/MP stay in my bag loaded with Spur Orthopan and slide film (Velvia 50) when I need or want the ultimate performance of the Leica lenses."

He has also written very favorably about the Zeiss ZM lenses. I guess he did as well as one could expect trying to make such a comparison. I have to wonder if Leica asked for the comparison? Although, I believe he is a truthful person, he is in a precarious position, not wanting to kill his golden goose! I think he usually reaches for "something for everyone approach", or seems to find something positive to remark. Like those who write for periodicals, they cannot totally trash all the products from their advertisers.

BTW-just what did you stir up?
happy.gif


Regards:

Gilbert
 
Cheers Gilbert!
happy.gif

Your quotes from Erwin are very interesting - I recall he was a big fan of Canon's 5D even using it as a sort of benchmark for DSLR performance. His comments about where the M8 fits in his kit seem very sensible and realistic - certainly it is doubtful that the M8 sensor can 100% exploit the capabilities of at least the most brilliant M lenses (135mm APO-Telyt; 50mm 'lux ASPH etc) especially when one of their most superb capabilities is their performance "in the field" which is generally unmatched in 35mm format.

Having read his views on the Zeiss offerings I have concluded that we 35mm rangefinder enthusiasts are indeed lucky that Zeiss had the balls to develop the Ikon and a range of excellent lenses - filling the mid-market gap and adding interest to the product segment overall.


Marc said: "I don't know what it is about Leicaphiles ... perhaps it's some form of delusional mental illness : -)"

I agree, too often I read comparisons between MF and Leica performance - this ignores the reality of "different horses for different courses". Maybe such comparisons are from those seeking to justify the small fortune spent on Leica M gear!
happy.gif


But Marc, take care...... because having "M8s" and God knows what other Leica M gear might just define you as a Leicaphile!
happy.gif
happy.gif
happy.gif


Paul said: "He kept telling how good they were for such a small negative.
I think these guys have got it all wrong."

I can hear him saying: "The photos look like shit, but gee for a mini-micro-nano-midget-dwarf sensor of 1mm x .05mm it is superb!"

I'm with you on that Paul. Fine to know a small format image is good and even to know how well it might compare to a larger format; but, by definition the comparison is generally useless - a smaller format just cannot outdo a larger format (ceteris paribus) and a direct challenge is meaningless - different horses....

Mark said: "Both cameras serve different purposes and this >type of comparison is meaningless to me."
AND
Paul said: "It would have been more useful to compare the M8 with the digital products from Canon and Nikon.
These cameras are being used for more or less the same fields of photography. "

Yep, generally I have to agree. What I would be interested in is a direct comparison of the M8 to the M7 loaded with film variations - e.g. Velvia; Neopan 400; 160NC; Neopan 1600; Provia 100 (oops the M8's lowest ISO is only 160 - that already poses a problem of comparison!) etc.. To my mind that would certainly be more meaningful.

Of course comparison with another DRF camera would be useful; but the only other DRF is the Epson and that comparison would only tell us what we already know! So other comparisons would need to be against DSLRs maybe of the same price bracket or sensor size or pixel count?

Personally the only way I would part with my M7 is if the image quality from the M8 was its equal - then a DRF could make sense to me!

And finally, this argument about 16 bit - well to my mind I could care less - how the sensor is "engineered" beyond it's size (has a direct impact on my shooting due to crop factor's impact on the focal lengths I use) and output, is techno-micro-detail I am disinterested in.

However, it reminds me that once my favourite Fuji Frontier lab people commented something along these lines: "16 bit is of little benefit in printing since the labs work in 8 bit". I hope I quote them accurately. But, if that is the case is it true that the benefits of 16 bit are lost at that point in the image chain? Are other digital labs able to operate in 16 bit?

Gee I'm now thinking that this is just the sort of crap that annoys me about digi-imaging - techno-mumbo-jumbo where geeks can froth at the mouth talking meaningless crap that ultimately just gets in the road of me enjoying taking photos - give me positive film any day - WYSIWYG!
happy.gif


Like Marc said: "I really don't need to know how the sausage is made, or want engineer anything ... I make photographs."

Finally, I am delighted that last weekend I got to see Hasselblad's 39mp back output - super posters printed on marvellous Ilford paper. Wow it was so impressive.
 
Hi Marc,

> Austin, you love to pour over how the sausage is made ... while > others just taste the sausage and decide if it's good or not.

Actually, I prefer both...but the problem is, knowing how the mechanism works, one tries to explain the results if they are different than what one expects. It's both a blessing and a curse.

I *always* encourage people to judge gear by the results, not by the specs. But, what sometimes happens when judging results, is the results don't tell the whole picture. No pun intended. For ex&le, the Contax N Digital. Straight out of the box, it wasn't at the top of the heap. But, once people learned how to use it...it gave some outstanding results.

That's also why there isn't just one camera available. Some criteria is subjective, and some is derived from needs. Everyone's needs and subjectivity is different. Obviously ;-)

Regards,

Austin
 
Yep, generally I have to agree. What I would be interested in is a direct comparison of the M8 to the M7 loaded with film variations - e.g. Velvia; Neopan 400; 160NC; Neopan 1600; Provia 100 (oops the M8's lowest ISO is only 160 - that already poses a problem of comparison!) etc.. To my mind that would certainly be more meaningful.>

Simon:

Better yet, your M7 and a couple of rolls of film and nice sunrise stroll.

Enjoy:

Gilbert

BTW-To me using a rangefinder is similar to sipping fine Scotch or Cognac!
 
Well, Well... You guys have thrown all the digital buzz words into the same black pot and cooked up a big mess. I'm in the know about this and will keep it simple and streight forward.

First off color bit depth has no overhead, I mean no ecc, no parity, etc. The following is how each pixel records color, it does not matter how many pixels you have it's the same process.

Each pixel is sensitive to light, red, blue or green so I'm going to talk about the green one but it works the same for all of them. The green pixel has 2 definite and opposite states white/green(on) and black/green(off). The bit depth between the two states will determine how good a picture looks. so basically it's the number of shades of green the pixel can record.

8 bit depth of color can record 256 different shades of green.
12 bit depth of color can record 4096 different shades of green
16 bit depth of color can record 65536 different shades of green
(Which box of crayons would you like to have?)

That said each RGB pixel of a camera with 8 bit color can record 16.7 million different colors but only 1 color per image. A camera with 12 bit color can record 68.7 Billion colors(computer geeks call this Hi-Color)and again only 1 color per image. A camera with 16 bit color can record 2,814 Trillion colors(computer geeks call this True-Color)and again only 1 color per image.

Beleive it or not a digital imaging sensor is an analog device(like film) and only after it is exposed to light is the information changed into it digital form. All this stuff about relevent data and overhead ECC, Parity, etc. comes into play with storage of the file and not with the bit depth of the analog to digital converter(the thing that takes the photograph and changes it to digital information).

What all this means is that the greater the bit depth the more sutulties in color change you will be able to record. these minute changes in color will really become apparent when you look at 2 pictures side by side but your eye will pick up the details in the shadows and the highlights even with out the comparison. Even if you're converting a 16 bit file to an 8 bit Srgb you will still see the difference in the print.

Now as for over or under exposing an image thats a whole other matter, but thats for another time.

I hope this has shed a bit(or 16)of light on the subject and I welcome any questions.

On my soapbox again!
Franc
 
Franc

Thank you very much for your good technical information .
It sounds very logical for me now , when you say the sensor itself is an analog device and true data , parity , ecc and control bits come into the game only after the the A/D conversion , when these data are on the way to the storage media .
 
Maybe we should have a "credentials" thread and get it over with once and for all : -)

Actually, it's not a bad idea since I personally like to know who is doing the talking about certain subjects.

When you think about it, those in the photographic arts are dependent on the word of those who make their tools more than ever before. They could make up anything, say anything, and I would just shrug my shoulders and say, "okay, whatever ... show me what it does". After all, my schooling and experience is in the visual arts, not in engineering.

And it is true that many, including me, parrot what they read or are told without a true understanding of all the engineering principles involved ... but belief is often based on the very thing we are discussing ... the credentials of the source mixed with first hand experience of the results.

Let's take the ex&le Austin used, the Contax N digital.

The promise was there based on the spec's. Full frame, 6 meg. (with large pixels), and a few questionable traits on paper. Okay, show me the results ... yikes! not so good. It didn't take an engineer to identify the problems, it took a bunch of photographers actually using the thing. Most of us determined that the software was the chief villain among others... we didn't need an engineer to determine that either.

The solution was to appeal to someone with credentials in software engineering for photography ...Thomas Knoll at Adobe. The minute Adobe included the N Digital among the cameras supported by Adobe Camera RAW, the Contax results substantially improved.
Unfortunately it was to late. That, and the other engineering flaws proved fatal.

Here's another ex&le I was also involved with first hand: The Leica M8:

I wasn't a beta tester for this camera, nor was I the first to get one. I WAS the first one to blow the whistle on the IR magenta issue. It took me one download to catch it, further test for it, and post the results on the Leica users forum ... creating a storm of "engineering babble" the like of which you wouldn't believe ... none of which served any purpose what-so-ever. The problem was in Leica's lap for them to sort out.

So Austin, I don't agree with your concept that if the results don't match the promise that WE have to know why. Maybe you do, but knowing why, isn't the same as fixing it ... that is the job of those who can actually do something about it. I don't want to discuss the solution, I want those who can FIX IT to discuss the solution.

Simon, as to the 16 bit issue verses printers being 8 bit: My understanding is that post work is best done with the most data possible going in. In almost every case a digital image is minuplated to some degree or another in a post processing program and masters of Photoshop as well as master printers have advised to stay in 16 bit mode for as long as possible... which Adobe has improved over the years by making more and more of their retouching tools work in 16 bit mode. As Franc implied, it doesn't take a genius to see the difference in a print that originated and was post processed as a 16 bit file verses a 8 bit one ... even printed on an 8 bit printer.

Whether all of that is true or not is irrevelant to me. What I do know is what imaging experts tell me to do to get better prints ... and in most every case they have been right based on the visable results I get using their advice. I now maintain 16 bit for as long as I can, I use 360 ppi with my Epson printers because a world famous master printer said so. And I now use a RIP to print through ... which, when combined with the other advice, has been a quantum leap forward in improving my inkjet prints.

In the few days we have been discussing all of this, I have shot five jobs and processed over 1,500 images to final printed form, printed forty 17"X22" display prints ... half of them from film scans, and over one hundred and fifty 8"X10"s.

... including some frivilous spring shots with the 39 meg. H3D just for fun : -)


26931.jpg
 
Yeaaaaaay.

It took just 60 posts to get to a PICTURE from an H3D 39 - on a thread called H3D.

Nice one, too, Marc. Love the edges of the plant. Plenty of detail there - 2814 trillion or so colors by the look of it. Of course I have no idea how many colors my brain is interpreting - but it sure looks good from this side of my eyeballs. ;-)

Cheers, Colin

PS. Looks just like my garden ... I wish.
 
Here's another H3D/39 shot using the HC 50/3.5 and a LEE grad filter .

These images are actually a paying job for a Cemetery ... a print c&aign to promote the idea that a Cemetery is for the living to visit in memory of loved ones ... and is kept beautiful for their enjoyment. It's an on going, year long photo project.

This pic really suffers from web compression because of all the blossom detail, but is incredibly detailed in the 17"X22" print.


26934.jpg
 
Hi Franc,

One thing that's missing from your post, IMO, is how the color depth/bit depth is used, and therefore it's significance to the discussion.

The human eye can't distinguish 16 bits of color depth, nor can any typical printer (from my understanding) print it, nor any typical monitor display it. The colorspace used may be 16 bits per color, but that is for the reason I mentioned...so you have headroom for color manipulations, but that does not mean all the bits are valid.

The most I have read that the human eye can distinguish is about 10 million colors max under one specific lighting condition. That would mean 8 bits per color, or 24 bits per color would be more than sufficient. So, no matter whether it can be recorded, we can't see it.

With respect to monitors, the digital part of the DVI spec (at least the most recent spec I have) provides for 24 bits per pixel, or 8 bits per color. CameraLink (which is almost identical to DVI, but for video capture instead of display), provides for 36 bit RGB or a max of 12 bits per color. Again, though you may use a 48 bit color space, that does not mean the monitor can display it. So, even if the data were in the image file, it could not be displayed on a typical monitor, much less be visible to us.

With respect to printers, that is one area I have only limited personal experience with current printers. Technically, for inkjet printers, the number of colors is more so limited to the size of your "dither", and what reduces the size of the dither is the number of inks you have and the size of the droplet...but from my experience and from many discussions on digital printing lists/forums that I've been on, my understanding is they typically don't use more than 8 bits/color or 24 bits, and they can't print near that many. More on the order of a few thousand. If someone has more information on this, I'd certainly like to know.

With respect to ECC/Parity etc., you're spot on. The only place there is typically ECC is in a memory subsystem or a disk subsystem. Parity can also be used in a memory subsystem or a disk subsystem, and is also included in *some* "transport" mechanisms, like SCSI. But, when used in a transport mechanism, it is not typically stored at the receiving end, only used to determine if there is an error.

An image file *may* have a "checksum" or a CRC, but that is on a per file basis, not on a per pixel basis.

> A camera with 16 bit color can record 2,814 Trillion colors(computer geeks > call this True-Color)and again only 1 color per image.

But, back to the initial point I was making, I still don't believe that the Hasselblad (or any commercial digital camera to date) actually produces 16 "good" bits. They may use a 16 bit word to store the 12/13 whatever bits, but I still contend all 16 bits are not good.

The only way to conclude that for a fact, is to get the spec sheet for the sensor (since it can't be better than the sensor), or do actual testing. I'll do a little research and see what I can find out and report back.

Regards,

Austin
 
One more from the Cemetery series that I shot this Sunday. The models could only come at 11AM ... worst light concievable and a true test of a digital camera's dynamic range ... or whatever the correct term may be : -) Sun just blasting away.

H3D/39, HC 50/3.5 ... Hensel Porty battery driven strobe at 600 watts to balance the shadows ... so windy that the diffuser had to be hand held or the whole thing would've taken off like a kite, sand bags and all.

This is a 30% crop of the original which will be used this way as well as the full shot with the little girl's family in the background. Which illustrates why clients want huge digital files to work with.

26937.jpg
 
Marc:

Great stuff with that young child. The 50 gave you a nice tight DOF - and the foremost tulips don't interfere because of the excellent placement of the face in the crop. The colors/shades held in the white tulips III to VII (?) are very good. Nice.

Cheers, Colin

PS Cemetery huh ... I take that back about wishing it was my garden
biggrin.gif
 
Hi,

Marc, don't let me interrupt your gorgeous pictures! Please, keep posting them! It dilutes the digressions ;-)

I believe the sensor in the 39m back is a Kodak KAF-39000. I got the spec sheet for it. The spec sheet clearly states:

"Linear Dynamic Range ... 71.4dB"

This translates into 12 bits of usable data. How that works is using a standard power conversion formula:

20 log (2**12) = 72.24dB

So, there you have it. Q.G.D. ;-)

The dynamic range is determined by the saturation and the noise level using again, a standard power conversion formula:

20 log(Vsat/Vn)

Vn = NR * Q/V

Q/V = 26uV/e Ne = 60k e

Vsat = Ne * Q/V = 1560000 uV

NR = 16 e

Vn = 16 e * 26uV/e = 416 uv

20 log(1560000uV / 416uV) = 71.4dB

Again, Q.G.D. ;-)

So, according to the Kodak spec, if the Hasselblad back uses this sensor (which I have confirmation from Kodak that Hasselblad does use this sensor), it's only giving you 12 bits of usable data. And, IMO, there is nothing wrong with that, it's way more than you can use.

The spec for this sensor is readily available on-line:

http://www.kodak.com/ezpres/business/ccd/global/plugins/acrobat/en/datasheet /fullframe/KAF-39000LongSpec.pdf

So, those who care to, and are capable of, can review my findings and conclusions.

Regards,

Austin
 
Marc, Fantastic images, Thanks for the trillions of colors in this form, it's very refreshing.

Austin, IMO it's like the audio digital vs analog debate. You don't have to be able to measure it to hear it. Although you cannot distinguish that many different colors you can still see a difference.

To your point about usable bits, I would tend to disagree, with 16 bit depth of color all 16 bit are relevent and in a properly exposed photograph you have every pixel recording relevent data. The camera manufacturers have made it easy to make sure all your data is relevent and the device we have is the clipping indicator. If 2 or more bits of the 8,12 or 16 bits at the bright end of the scale are the same, the clipping indicator will show you on your preview screen. Exposure is even more critical with digital, thats why the people who have experiance in shooting chrome film take to digital like ducks to water and the neg guys like me have a bit of a challange. Digital has little latitude 1/5th stop over exposed and 1/3 stop under exposed, where as I could still get a good printable image even one stop out on negative film.

But my basic point is this, If you properly expose your image you will get TRUE 16 bit depth and you will see the difference in the image.

Just my nickles worth,
Franc Flipsen
Electronics Engineering Technologist,
Micro Computer Repair Specialist
 
You must log in or register to reply here.
Back
Top