PDA

View Full Version : Upsizing R-D1 files. G'Fractals, etc.?


SimonPJ
04-18-2005, 01:42
Following Sean's advice about avoiding the default clipping in the raw convertors (Epson and PS) I've found some useful improvements in getting shadows to hold up.

But... I still find that for much of my Leica-style available light shooting I get poor, noisy transitions into shadow areas that are quite noticeable on A4 or larger prints.

At 14x11 I find that head and shoulders portraits have poor definition in the hair (especially darker hair) - even with ideal original files and post-processing.

Do programs such as Genuine Fractals work for these kinds of problems?

Thanks,

Simon.

Sean Reid
04-18-2005, 04:36
Hi Simon,

No, unfortunately, Fractals tends to give the files a kind of waxy softness which actually deteriorates fine detail. Fractals prints look great at a distance but their sort of painterly quality is obvious close up (to me at least).

Cheers,

Sean

kbg32
04-18-2005, 06:17
I find that giving my files a slight sharpening in PS CS unsharp mask before printing, really helps the final print output.

Amount - 300
Radius - 0.3
Treshold - 0.

kbg32
04-18-2005, 06:21
As for upsizing in CS, use bicubic smoother. I have an Epson 2200. Even though I print at 1440, you don't need to go beyond 300 dpi in the final file size. You will never notice the difference, even in a 13" X 19" print.

J. Borger
04-18-2005, 07:02
Simon .. there are too many parameters to consider here. Exposure, Raw- conversion, processing in ps, lens used, printer, paper.....
For 11X 14 you do not need a program like GF. You can do it in PS and i suggest to consider printing at 240 - 260 dpi instead of the usual 300 dpi to upsizing first.... !
The key to the problems you describe are: good exposure, raw-processing, processing in PS and appropriate amount of sharpening.
Besides results may differ due to paper/ printer combinations.
If i print the same B&W image on an Epson 2200 on matte vs semi glossy paper for instance there is a huge difference in detail in the black & shaded areas.

I do not think GF or the way you upsize images is a solution to your problems.
No matter how you upsize: the results at 14 x 11 should be flawless!!

Han

Sean Reid
04-18-2005, 07:10
Agreed on 240. In the tests I've done, Epson printers like to be fed data at 240 or 480 ppi (not dpi which is a print measure of course). There's only a very tiny advantage to using 480 ppi and I print at 240 ppi almost exclusively.

Cheers,

Sean

J. Borger
04-18-2005, 08:01
Thanks for correcting on the PPI Sean.
The 480 PPI is new to me ... what would in short be the reason to prefer 240 ppi to 300 PPI?
What i basicaly do is resize and print at the PPI i get ..... as long as it does not go below ca. 240?

BTW any tips on paper for B& W printing on the Epson 2200 with the Harrington Rip?
I print on Epson enhanced Matte and Semi glossy so far.... stopped with the semi glossy because the bronzing is killing me. Is there a paper a bit less matte than Epson Matte ... let's say with a surface like a page in a photobook?

Han

Sean Reid
04-18-2005, 08:29
To be honest, it's been so long since I first looked into (and decided upon) output res. settings that I can't remember the details. When photographers first started using these Epson printers for serious photo work there was a list run by Mitch Leben that included many very knowledgeable people and we were all making tests and sharing results. I can't recall why this is but Epsons' drivers like to get their data at 240 ppi or a multiple thereof. Try a print and you'll see that 240 will look at least as good as 300. 480 shows just a tiny jump in apparent sharpness but only if the original files was large enough to print at 480 without interpolation. Upsizing a files to print at 480 wouldn't gain one anything and might actually degrade the print. If I'm sending for a book, brochure, magazine, etc. I send it at 300 ppi. I send files to my 2200 at 240 ppi with the Epson and Harrignton drivers. What does Harrington recommend? I can't recall.

I proof on enhanced matte and am still experimenting on equivalent papers for archival prints. Try Epson Smooth. I'll try to post more on this later as I decide further and please keep us posted on any papers you discover.

Cheers,

Sean

Sean Reid
04-18-2005, 08:31
To be honest, it's been so long since I first looked into (and decided upon) output res. settings that I can't remember the details. When photographers first started using these Epson printers for serious photo work there was a list run by Mitch Leben that included many very knowledgeable people and we were all making tests and sharing results. I can't recall why this is but Epsons' drivers like to get their data at 240 ppi or a multiple thereof. Try a print and you'll see that 240 will look at least as good as 300. 480 shows just a tiny jump in apparent sharpness but only if the original file was large enough to print at 480 without interpolation. Upsizing a files to print at 480 wouldn't gain one anything and would likely degrade the print. If I'm sending a file for a book, brochure, magazine, etc. I send it at 300 ppi. In house, I send files to my 2200 at 240 ppi with the Epson and Harrignton drivers. What does Harrington recommend? I can't recall.

I proof on enhanced matte and am still experimenting on equivalent papers for archival prints. Try Epson Smooth. I'll try to post more on this later as I decide further and please keep us posted on any papers you discover.

Cheers,

Sean

SimonPJ
04-18-2005, 10:28
Thanks for the helpful replies. Using 240ppi will obviously make a difference for upsizing files.

I'm still not convinced that what I'm coming up against aren't the resolution and dynamic range limitations of the camera (chip, processing, etc.) But I'm hoping to find that I'm wrong! I've certainly got a lot to learn about digital image processing.

Most of my shooting is of people going about their business indoors in mixed lighting, usually from the dark side, and often with bright window highlights behind. I use incident readings from the dark side for that situation - which usually means that I'm giving several stops more exposure than the R-D1 would if set to automatic.

I have shot colour negative and the R-D1 together in the same situations, at the same speeds (400 or 800) and with identical exposures (often 125th or 60th at F2). I don't have any problems getting good shadow detail and tonality from my film/digital workflow, using a Minolta 5400 scanner, PS CS and Epson 1280 at up to 14x11.

But I do find that the R-D1 prints, which look ok at first glance, often have ugly transitions into the shadows on the fleshtones. Where this is noticeable is as the light falls away across a cheek, or where hair casts a shadow on a face - or in the modelling from light to shade across an arm. With film I have fewer problems getting gentle transitions that read across the flesh contours. But with the R-D1 I'm finding it much harder to get the same results when shooting in the same way. Detail and range of tonality in dark or shadowed hair are also a problem.

I usually don't pick up the problems on screen - it's only on the prints that its obvious.

I don't think I'm underexposing the R-D1 files. The histograms look ok, without obvious clipping (except in highlights that I don't care about). I obviously need to work harder at extending the range of shadow tonality in post-processing.

At this stage of my experience with the R-D1 it feels that for the kind of documentary-style shooting I do most of the time, with variable and poor interior lighting, often of the shadow side, I'm finding the same kinds of limitation that I'd expect if I was shooting with fast slide film - compounded by some resolution limitations.

I hope I can start to prove myself wrong....

Simon.

pfogle
04-18-2005, 12:39
Simon, sorry if I missed something, but are you using 48 bit files? Seems necessary here for the shadows.

Phil

jlw
04-18-2005, 13:00
I've also noted abrupt transitions in shadowed skintone areas when printing on an Epson 1280 from both digital images and scans. I think it's more a printer issue than a file issue, although scans from film may be "noising up" the transition areas enough that they don't look as abrupt.

If it makes you feel any better, my current Epson R800 does even worse than the 1280. Maybe the answer is some kind of trick RIP or custom inkset (or one of those HP printers with a separate set of grayscale inks...)

SimonPJ
04-18-2005, 13:24
Yes, Phil - I'm using 48 bit files. I suppose the printer may be the problem, as jlw suggests - though it does work much better with my scans. Maybe, as jlw suggests, it likes the noise characteristics of the scans better.

But at the moment I still feel that the difference is about more than that...