Foveon reds re-visited - a Fix

[xpatUSA]

Having displayed some recalcitrance in a previous thread on Foveon color rendition, see here, it's time to eat some humble pie and to thank all those gentlemen such as Colin Southern who all mostly pointed a finger in the right direction.

I had played the "I'm only a watch hobbyist" card to the hilt and blatantly declared no intention of upgrading software, changing my lighting, etc, ad naus. I even sold the Sigma SD9 in a fit of pique. Well, I've been regretting that particular move.

So:

I now have Camera Raw 5.4 installed (the latest possible with PSE6). That brings the ability to select DNG camera profiles for an image or a set thereof.

I've downloaded Adobe's free RAW to DNG converter.

I've downloaded Adobe's free DNG profile editor with it's sliders and such.

I've ordered a Macbeth mini card.

I've returned the newly purchased Nikon D90

I've bought a Sigma SD10 on eBay

A steep learning curve lies ahead but "Cinderella, you shall go to the Ball . ."

Thanks to All,

Ted

[Colin Southern]

No worried Ted.

As a "born again" starting point, why not see if someone already has a DNG profile for a SD10 - it'll be a LOT easier than using the DNG profile editor. (or splash out another $100 and get a Color Passport) (worth their weight in gold).

I still don't believe that the SD10 will be any sharper than a camera with a conventional sensor though, and capture sharpening pretty much compensates 100% for the bayer sensor and anti-aliasing filter.

[xpatUSA]

No worried Ted.

As a "born again" starting point, why not see if someone already has a DNG profile for a SD10 - it'll be a LOT easier than using the DNG profile editor. (or splash out another $100 and get a Color Passport) (worth their weight in gold).

I still don't believe that the SD10 will be any sharper than a camera with a conventional sensor though, and capture sharpening pretty much compensates 100% for the bayer sensor and anti-aliasing filter.

Thanks Colin, more good advice, thanks again.

The Macbeth mini-card arrived today it's perhaps the key to using the Profile Editor. Take a pic of the card - convert the .nef to a .dng - open the .dng in the Profile Editor and go the colorchecker tab where it examines the card image and makes a profile automatically - which can then be tweaked as one thinks fit.

On sharpness, I can only offer my own experience: a SD9 vs. D50 test conducted in a statistically insignificant manner (one sample of each camera, only a few shots taken, different lenses). I took SD9 and D50 raw pics of an ISO target at about 4 ft distance. Converted the RAW files to 16-bit TIFF and examined each one with QuickMTF:




No sharpening applied to either image. The D50 had the 60mm micro-Mikkor, the SD9 had a 50mm EX but pre-DG macro lens on it. Manual focusing was used. It is clear from the graphs which camera is the sharper insofar as edge spread function is concerned. I'm not saying that this proves anything (too unscientific) but my experience does follow the concensus in the literature.

Armed with the SD10 and my new Sigma 70mm EX DG f/2.8 macro, I'll probably try another shoot-out sometime in future.

We shouldn't pixel peep but, with 9.12um pixel pitch every pixel counts:



Can we guess which camera has the Bayer sensor :-)

all the best,

Ted

[Colin Southern]

No sharpening applied to either image. The D50 had the 60mm micro-Mikkor, the SD9 had a 50mm EX but pre-DG macro lens on it. Manual focusing was used. It is clear from the graphs which camera is the sharper insofar as edge spread function is concerned. I'm not saying that this proves anything (too unscientific) but my experience does follow the concensus in the literature.

Ted

Hi Ted,

The experiments are one thing -- but real world photography is something different. As a good example; the Bugatti Veyron is good for somehting in excess of 400km/hr (250 MPH+), whereas the Ferrari Enzo may only be good for (best guess) 350km/hr. So clearly the Veyron is the better car right? Technically perhaps, but if it's only needed for a shopping trip down to the local supermarket then one could (I feel) argue that either are "more than adequate".

"More than adequate" is a term that we don't hear very often in photographic circles ... and in my opinion we SHOULD hear it far more often; for want of a better phrase, what I'm seeing is an "over focus" on things that folks think that they understand (that marketers take full advantage of) but not realising that it doesn't translate through to real world applications the same. Example; when I had an 8MP 20D I used to throw away 88% of the data when I made a typical image to be displayed online. Now that I have a 21MP camera, I'm throwing away 95% of the data. When it comes to pixels, the 8MP was more than adequate most of the time. Same with bayer -v- Foveon sensors I'd suggest; any bets that I couldn't take a closeup photo of my watch with my camera (which has a Bayer sensor) that wouldn't knock your socks off in terms of detail?

[xpatUSA]

Hi Ted,

"More than adequate" is a term that we don't hear very often in photographic circles ... and in my opinion we SHOULD hear it far more often; for want of a better phrase, what I'm seeing is an "over focus" on things that folks think that they understand (that marketers take full advantage of) but not realising that it doesn't translate through to real world applications the same. Example; when I had an 8MP 20D I used to throw away 88% of the data when I made a typical image to be displayed online. Now that I have a 21MP camera, I'm throwing away 95% of the data. When it comes to pixels, the 8MP was more than adequate most of the time. Same with bayer -v- Foveon sensors I'd suggest; any bets that I couldn't take a closeup photo of my watch with my camera (which has a Bayer sensor) that wouldn't knock your socks off in terms of detail?

Hello Colin,

Well, most of my watch pics are of the whole watch, say 40mm tall on an APS-C sensor, approx 1:4, so not much "detail" per se would be seen in my typically 640 to 800 px images.Therefore, a 3MP camera does just fine for that. 21MP does sound like a lot of resolution power and more than adequate for such a watch picture. I would imagine that the necessary re-sampling from a huge 21MP down to an 800px wide image would take care of all that Bayer-induced uncertainty, by courtesy of the power of the in-camera or post- processing.

Meanwhile, would love to see a pic of your watch!

Ted

[Colin Southern]

Hi Ted,

I'll take one after dinner if I remember.

[Colin Southern]

Here you go Ted.

This is a medium resolution shot (be sure to click on it to se it 1200 x 1200px) - how'd I do?

[xpatUSA]

Very nice, Colin!

A popular watch - I never knew they had "Rolex" written all round that ring under the crystal. EOS 1Ds Mark III, eh? Quite a camera - what was the lens you used?

I would like to reduce your copyrighted image to 1/2 size (638px, more in my realm) and keep it as standard by which to judge the SD10 results.

best,

Ted

[Donald]

... what was the lens you used?

Ted

The EXIF data shows us he had the TS-E90mm f/2.8 on board for this one.

[xpatUSA]

Ted

The EXIF data shows us he had the TS-E90mm f/2.8 on board for this one.

Thanks Donald,

I missed that, not used to looking through EXIF stuff!

Ted

[rpcrowe]

Ted, I strongly recommend that you upgrade your PSE-6 to PSE-10. PSE-10 can be had at a relatively low cost but, is far superior to PSE-6. PSE-8 was a big upgrade over PSE-6 in the Elements series and PSE-10 took it further in closely matching the capabilities of their big brother the Photoshop CS Series. http://prodesigntools.com/photoshop-...n-compare.html

I never actualy used PSE-6 however, when I upgraded to PSE-10 from PSE-8; I gave the PSE-8 to a friend who was, at the time, using PSE-6. She tells me that 8 is a quantum improvement over 6.

[Colin Southern]

Very nice, Colin!

Thought you might like it

A popular watch - I never knew they had "Rolex" written all round that ring under the crystal. EOS 1Ds Mark III, eh? Quite a camera - what was the lens you used?

Yes - they put the serial number there too (makes it kinda hard to remove). It's a pretty old camera now to be honest, but any modern camera would have given pretty much the same result. The lens was (as Donald correctly identified) at TS-E 90; it's my standard lens for art reproduction. It gets in pretty close, but not close enough for work like this, so I used an extension tube (it's really the realm of the macro lens, but I don't own one).

I would like to reduce your copyrighted image to 1/2 size (638px, more in my realm) and keep it as standard by which to judge the SD10 results.

Yep - that's fine. Jest remember though that it's not about the camera (it only captures the image) - it's the sharpening and retouching that's responsible for 90% of the finished quality (although getting a high-quality capture to start with certainly helps) (I think I used something like $20,000 worth of equipment to light and capture the image!).

[Colin Southern]

PS: Here's a link to the full-resolution file (3.5MB)

http://backup.cambridgeincolour.com/www.pbase.com/cjsouthern/image/143093109/original.jpg

[xpatUSA]

Colin,

Thanks for the original image, duly downloaded. The Sub bezel diameter, is it 40mm? I'm currently figuring out what detail I might expect from a larger pixel pitched camera than yours. (D50: 7.8 um, SD10: 9.12um).

Ted, I strongly recommend that you upgrade your PSE-6 to PSE-10. PSE-10 can be had at a relatively low cost but, is far superior to PSE-6. PSE-8 was a big upgrade over PSE-6 in the Elements series and PSE-10 took it further in closely matching the capabilities of their big brother the Photoshop CS Series. http://prodesigntools.com/photoshop-...n-compare.html

I never actualy used PSE-6 however, when I upgraded to PSE-10 from PSE-8; I gave the PSE-8 to a friend who was, at the time, using PSE-6. She tells me that 8 is a quantum improvement over 6.

Hi Richard,

I ended up with PSE6 after a long battle with the Organizer in PSE7. The Organizer was totally obtrusive and I got sick of having to uncheck the stupid "include in the Organizer" checkbox every time I saved an image. It seemed that I was not alone, because there was plenty of advice on the 'net about how to kill it. Tried that but I ended up with the world's most sluggish PSE7 and no desire to download it again over my 56K internet connection. I looked at PSE8 at the time and all I could see was more bloatware and functions that I would never use, or so I thought. So I bought a PSE6 CD for $20 on eBay and have had no problems with it to date. I've been able to re-load it up after several computer crashes and changes of model. I'm mindful though that PSE10 will likely bring a later version of ACR with more color matching stuff. (I'm on ACR 5.4 which is as high as gets with PSE6).

I'll certainly check out your link and I do appreciate your advice. What, in your opinion, would PSE10 bring to an amateur table-top photographer? That was a serious question, not rhetorical, by the way.

thanks,

Ted

[Colin Southern]

The Sub bezel diameter, is it 40mm? I'm currently figuring out what detail I might expect from a larger pixel pitched camera than yours. (D50: 7.8 um, SD10: 9.12um).

Hi Ted,

It's 30mm across the black face of the watch (ie the diameter of the ring that has "Rolex and the serial number on it"). The camera won't make a great difference though; just get as close as you can with a macro lens - light it properly - and sharpen it properly (in addition to other normal retouching) and you should be just fine.

[xpatUSA]

Hi Ted,

It's 30mm across the black face of the watch (ie the diameter of the ring that has "Rolex and the serial number on it"). The camera won't make a great difference though; just get as close as you can with a macro lens - light it properly - and sharpen it properly (in addition to other normal retouching) and you should be just fine.

Hello Colin,

Thanks for the advice on how to take pictures of watches.

I should have explained more clearly that, by "bezel" I meant the outside diameter of the rotating part. For a Sub's "bezel diameter" a bit of googling gives 40mm but I just wanted to check.

On second thoughts, it would be a little difficult to use your image as a comparison standard. Closer examination of your original large image showed it to be in gray-scale, processed such that it has an incredible MTF of 50% at just under Nyquist and an almost perfect edge rise 10-90% of 1.17 pixels. A veritable tour-de-force of macro imaging!

On the other hand, I myself normally make performance comparisons between images with no processing at all, apart from PSE/ACR's conversion to 8-bit RGB with everything set to 0. This restricts a comparison of acutance between cameras to the more mundane parameters of manual focusing ability, correct aperture, lens performance, sensor performance and so forth.

I mention MTF only because there was a handy-dandy edge in your original image which looked pretty sharp and I had the means to measure it. MTF for most of us can be terribly misleading and thereby hard to interpret. My first pic with the now-departed D90 was a hand-held shot of an A4 sized ISO target from about 4 feet away. The MTF curve was horrible! Not surprising, though. After mucho post-processing and re-sampling to web size, the MTF became quite respectable!

best regards,

Ted

[Colin Southern]

Hello Colin,

Hello Ted

Thanks for the advice on how to take pictures of watches.

You're very welcome.

I should have explained more clearly that, by "bezel" I meant the outside diameter of the rotating part. For a Sub's "bezel diameter" a bit of googling gives 40mm but I just wanted to check.

Yes, it's 40mm

On second thoughts, it would be a little difficult to use your image as a comparison standard. Closer examination of your original large image showed it to be in gray-scale, processed such that it has an incredible MTF of 50% at just under Nyquist and an almost perfect edge rise 10-90% of 1.17 pixels. A veritable tour-de-force of macro imaging!

Since it's a "black and white" watch I simply set saturation to zero because I was too lazy to white balance it correctly! I also set it to grayscale later on in an attempt to get the full resolution file size down some more. Measurements of things like MTF etc are totally meaningless as the image has been retouched, including the extensive use of burning shadows and dodging highlights to force the contrast.

On the other hand, I myself normally make performance comparisons between images with no processing at all, apart from PSE/ACR's conversion to 8-bit RGB with everything set to 0. This restricts a comparison of acutance between cameras to the more mundane parameters of manual focusing ability, correct aperture, lens performance, sensor performance and so forth.

I mention MTF only because there was a handy-dandy edge in your original image which looked pretty sharp and I had the means to measure it. MTF for most of us can be terribly misleading and thereby hard to interpret. My first pic with the now-departed D90 was a hand-held shot of an A4 sized ISO target from about 4 feet away. The MTF curve was horrible! Not surprising, though. After mucho post-processing and re-sampling to web size, the MTF became quite respectable!

Arrrggghhh. Ted - forget all these things; they won't make any visible difference. It's like comparing chrome-vanadium coated ball-bearings -v- regular hardened steel in a tour-de-France racing bike. Put Lance Armstrong on either of them and he's still going to finish the course about 18 months before me, because it's not about the bike. In photographic terms, give me an old Canon 300D and I'll bet that I could shoot a model in my studio - publish the results here - and not a single soul would realise that I wasn't using my usual 1Ds3, because it's not about the camera.

[xpatUSA]

Arrrggghhh. Ted - forget all these things; they won't make any visible difference. It's like comparing chrome-vanadium coated ball-bearings -v- regular hardened steel in a tour-de-France racing bike. Put Lance Armstrong on either of them and he's still going to finish the course about 18 months before me, because it's not about the bike. In photographic terms, give me an old Canon 300D and I'll bet that I could shoot a model in my studio - publish the results here - and not a single soul would realise that I wasn't using my usual 1Ds3, because it's not about the camera.

Colin,I can't do that, it's a sad personality trait. I have to know how something works down the finest detail. Can't help it.Yes, I am well aware of the philosophy and I do subscribe to it but in my own way. Hence the purchase of a 3MP Sigma SD10 for web shots of watches and the return of the newly-acquired Nikon D90 when I realized that the D50 is perfectly adequate for pointing-and-shooting around here. And as for my lighting . . .

To illustrate further, none of my watches cost me more than about $700 and my daily "beater" watch is a restored and slightly customized military plastic watch. Unfortunately, that doesn't stop me posting highly technical stuff on a couple of watch fora, where I have acquired a rep. for being somewhat pedantic. I construct rather plain-looking wooden buildings while doing my own stress calcs and researching architectural style. Also, I play a home-built guitar and a Yamaha Strat copy through a really old and beat-up amplifier . . . but delve deeply into music theory especially that of harmony - while no longer being in live entertainment and jamming with my buddy once a month, if that. At 72, it pays to keep the brain active!

However, I will try (trusssst me) to refrain from using technical arguments on this forum where "form is over function" so to speak

[Colin Southern]

Colin,I can't do that, it's a sad personality trait. I have to know how something works down the finest detail. Can't help it.Yes, I am well aware of the philosophy and I do subscribe to it but in my own way. Hence the purchase of a 3MP Sigma SD10 for web shots of watches and the return of the newly-acquired Nikon D90 when I realized that the D50 is perfectly adequate for pointing-and-shooting around here. And as for my lighting . . .

To illustrate further, none of my watches cost me more than about $700 and my daily "beater" watch is a restored and slightly customized military plastic watch. Unfortunately, that doesn't stop me posting highly technical stuff on a couple of watch fora, where I have acquired a rep. for being somewhat pedantic. I construct rather plain-looking wooden buildings while doing my own stress calcs and researching architectural style. Also, I play a home-built guitar and a Yamaha Strat copy through a really old and beat-up amplifier . . . but delve deeply into music theory especially that of harmony - while no longer being in live entertainment and jamming with my buddy once a month, if that. At 72, it pays to keep the brain active!

However, I will try (trusssst me) to refrain from using technical arguments on this forum where "form is over function" so to speak

Hi Ted,

The big issue as I see it is that there are many other parts to the formula ("variables" if you like) that need to be considered (but you're not considering). Working out MTFs is one thing, but if you really want to do it "by the numbers" then you'll need to add things like sharpening - monitor resolution - viewing distance from the monitor - magnification - how good our eyes are etc into the mix.

[xpatUSA]

Hi Ted,

The big issue as I see it is that there are many other parts to the formula ("variables" if you like) that need to be considered (but you're not considering). Working out MTFs is one thing, but if you really want to do it "by the numbers" then you'll need to add things like sharpening - monitor resolution - viewing distance from the monitor - magnification - how good our eyes are etc into the mix.

Colin,

It has just become a non-issue as far as I'm concerned. It is difficult to accept that my literary style has given the impression that I know so little about photography but that is how it seems, judging from the above advice. So, I'll just bow out of this ill-fated thread. Take care,

Ted.