Camera megapixel ratings - are we being fooled

[Puffer Fish]

Damn Puffer Fish you available for hire to teach classes? (If you don't already)

Thank you - I give classes in color theory, how humans see color and technical measurement of color - but working with images was necessary to understand how to duplicate real objects. I get paid to understand all of this as part of my job. Anyone who want information - that is free.

Issues like Gamut mapping and image translation and preserving dynamic range tend to either get lost or forgotten about. The world of imaging between film and digital is so different, that it is easy to assume things are similar when they are not.

If anyone wants details - specific information or proof, all they have to do is ask.

My spectophotometer budget is a little over $450,000, so some of what I get to work with is outside the normal person's pocket book.

 

[ScubaLuke]

SCUBALUKE- consider the following

1. The sensor size on the D2X is 23.7 x 15.7, or 372 square mm. The number of pixels is 12.8 meg. That is a sensor resolution of 34408 pixels per sq mm, or 185.5 lines per mm. Any lense out there capable of that???

Actually, it would be at 92.75 lines per mm. You need two pixels to resolve a line. But that still pushes the limits of most lenses. The issue for me is mainly one of cost. At the moment the D2X is going for around $4500 from B&H and then I still have to go get a housing. And the last time I checked Light and Motion still hadn't signed up to a D2X housing and I love my L&M Stingray housing.

3. D200 is a far better deal, by the way and has almost the identical sensor size (but only 165 lines per mm)

True if I were going to change systems now, that's probably what I would get. But I still run into the having to buy a new housing problem.

Want better looking digital images (your camera is not lens limited, but you have to do some work to get the best quality - similar to film, just really, really different).

Suggested handling methods:

1. Shoot in RAW - giant files, but you now actually have the actual data the sensor captured.

That's something I hadn't started doing until recently and I haven't done any UW work in raw yet.

5. Get a very high resolution printer, but you have to understand that in the digital world, every change looses data. If your image is in 350 dpi and the printer prints in 233 for example, the machine has to translate (loosing lots of information in the process). This gets really tricky because the printer company will not tell you exactly what they are really printing in, or how their software does the translation. A 8 color printer may actually only be printing with 5 head, 6 head or 7 heads. The actual dpi, if stated at say 2800 could be 560, 466 or 400 and you will have no way of knowing. What is important is to have matching resolution in some ratio 1:1, or 2:1 or 1:2 for example. Anything else requires data to be lost, and it can be a lot.

6. If the image size and the printer image size does not match, then recalculate the image into one that does and look at the effect before printing.

I've got an older Epson 1280 printer which was hot stuff for it's day but there's a lot better stuff out now. I'm thinking of going to one of the online printing services. Would you go for one of them or a better printer?
Luke

 

[Puffer Fish]

Actually, it would be at 92.75 lines per mm. You need two pixels to resolve a line. But that still pushes the limits of most lenses. The issue for me is mainly one of cost. At the moment the D2X is going for around $4500 from B&H and then I still have to go get a housing. And the last time I checked Light and Motion still hadn't signed up to a D2X housing and I love my L&M Stingray housing.


True if I were going to change systems now, that's probably what I would get. But I still run into the having to buy a new housing problem.


That's something I hadn't started doing until recently and I haven't done any UW work in raw yet.


I've got an older Epson 1280 printer which was hot stuff for it's day but there's a lot better stuff out now. I'm thinking of going to one of the online printing services. Would you go for one of them or a better printer?
Luke

Thanks for correcting me on the lpmm - I have to admit that it gets really confusing because there are actually 3 sets of images (RGB), so I actually do not know how to do the total math correctly on that.

The printer is a difficult question. With Epson, the only really good choices in a smaller size are the R800 or R1800. Both can print finer that the human eye can resolve. These are not give away printers, so there is fair cost involved. If you buy commercial inkjet paper, you have about 25 cents per 8.5 x 11 prints, and about equal for the ink (obviously varies and could be more). That means around $.50 - $.70 per 8.5 x 11. Life expectancy on the prints is 3 - 5 times the best Kodak prints.

If you include the printer life of around 3,000 prints (again, might be more), The cost is around $.75 - $.85 per 8.5 x 11. I don't know how that compares or what specific technology the service would use.

What you get from having your own is the ability to learn a whole bunch quickly. Matching gamut's, match resolutions and understanding how everything works will only happen if you can see it for yourself. Everything you learn can directly apply to commercial printing, so knowing what relative gamut mapping is, for example has a actual advantage later (relative refers to relative, absolute or proportional mapping)

Get a printer and message me and I can give you directions for how to calibrate and do test printing.

 

[Xanthro]

Thank you - I give classes in color theory, how humans see color and technical measurement of color - but working with images was necessary to understand how to duplicate real objects. I get paid to understand all of this as part of my job. Anyone who want information - that is free.

Issues like Gamut mapping and image translation and preserving dynamic range tend to either get lost or forgotten about. The world of imaging between film and digital is so different, that it is easy to assume things are similar when they are not.

If anyone wants details - specific information or proof, all they have to do is ask.

My spectophotometer budget is a little over $450,000, so some of what I get to work with is outside the normal person's pocket book.

You have that rare ability to take complex issues and present the information in an understandable format, that both the novice and more techincally oriented can understand. Since message board posts tend to be written rather quickly, this is even more impressive.

As I gain in knowledge I'll take you up on the offer of answerning questions. I've just started with photography, so I am still just a novice. I understand color and light theory better than I understand the practical aspects. I hope to change that soon.

 

[hvulin]

Concerning the Corel vs Adobe... proce matters A LOT to me :-(
Has anyone used Gimp 2? Its main advantage is price (free) and platform (works great in Windows, but also in Linux.) Opinions?
Thanks in advance...

very good program (I use it in windows and linux) but with two issues:
1. doesn't support colour profiles for your devices
2. doesn't support raw

other than that it's the best you can get for 0$ (and much more)

 

[ScubaLuke]

What you get from having your own is the ability to learn a whole bunch quickly. Matching gamut's, match resolutions and understanding how everything works will only happen if you can see it for yourself. Everything you learn can directly apply to commercial printing, so knowing what relative gamut mapping is, for example has a actual advantage later (relative refers to relative, absolute or proportional mapping)

Get a printer and message me and I can give you directions for how to calibrate and do test printing.

Thanks,
I think that's good advice. It's really like doing your own darkroom work isn't it? The best results come from knowing your equipment and how it performs in different situations.
Luke

 

[markfm]

Very cool toys, PufferFish I don't get to play much with stills at work, just video streaming with long-range CCTV. ToyMaster is a fun role.

 

[Puffer Fish]

Thanks,
I think that's good advice. It's really like doing your own darkroom work isn't it? The best results come from knowing your equipment and how it performs in different situations.
Luke

Like a darkroom, only a 100 time better and you can correct your mistakes for free (unless you printed it). You can do things that no mire mortal should be able to -just takes some learning and practice. Compared to the cost of a darkroom, this stuff is cheap. Good luck.

As a side note, while I have and use adobe CS - I am far from an expert in it. It has way to many translation problems to be vary accurate. But I can help with Corel.

 

[Puffer Fish]

Very cool toys, PufferFish I don't get to play much with stills at work, just video streaming with long-range CCTV. ToyMaster is a fun role.

Fun and a bit frustrating. Technology is spreading, but are also both advancing and alienating people. Look at Tom's hardware site and their monitor reviews - they now use spectrophotometers, and then found out that they pretty much suck (one delta E is now excellent, when I need better than .3). Equipment makers are now getting better, but it will be years before a good one shows up that is affordable. The issue of mega pixels (this thread) and image quality are so full of junk and madeup concepts it is hard to separate truth from fiction.

A good example is Adobe CS, if professionals actually knew how pixel information was handled inside of that program, they would never use it. They don't, and it has been better than any other option, so everyone uses it.

On a side note, the people on this thread are really nice.

 

[markfm]

The straight tech info threads are all pretty good around here. People have opinions, exchange ideas, learn.
Back to the battleground -- MOFNMOF