I,m looking to buy the D1 very soon and was needing to know the best way to create a large, high quality file to produce a really good looking 16x20 or 20x24. I am a professional photographer just wanting to get into the digitel area and that is my main concern is what camera would be the best for making enlargements if needed or the best software I could use for this purpose. My main use for this system would be 8x10s or smaller,but from what I've seen they will all do that! What direction should I go?--Thanks Mr. Peeck,
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Creating really good 16x20 or 20x24 prints from the D1
- Thread starter Mr. Peeck
- Start date
L
Lin Evans
Guest
The software choice of most professionals for this type work is Genuine Fractals by Altamira. Basically what is being done is to take the native output from the camera and run it through a fractal interpolator. This mathmatical algorithm looks at each pixel and its adjacent pixels then makes a logical decision based on their values what additional pixels, if originally present would have looked like. It then actually creates these pixels and the result is a greatly increased file size which is suitable for printing at larger sizes.
The important consideration for your application would be how amenable the original data is to fractal interpolation. Because of hardware limitations (computer processor, RAM, and mass storage capacity) there have been some necessary limitations on the number of calculations which can be done in a reasonable period of time. This, in plain words, means that some images are more amenable to this process than others. Typically, an image with one or two large subjects would respond much better than an image with multiple smaller subjects. For example, a portrait would generally be quite amenable to this process while a group shot of 50 individuals taken in wide angle from a greater distance might tax the system.
You should consider the native aspect ratio of the camera when making these decisions. Check the ratios between horizontal pixel count and vertical pixel count. Then compare this to the desired output. If the ratios are identical, you will only need to enlarge to the degree that your print device requires. If the ratios differ, you may need to enlarge to a greater degree then crop the image to get the desired aspect ratio. Since most common print sizes are based on 35mm, a camera which most closely approximated the 35mm ratio of width/height would be best. Some of the newer 3.34 megapixel cameras offer a choice of two capture ratios. The Nikon CP990 is one of these as is the Sony DSC-S70.
The D1 appears to be an excellent choice, though you MAY need to spend a little time in the digital darkroom to get the color balance and flesh tones like you want. On the other hand, there are excellent software packages (QimagePro and Bibble come to mind) which have been able to correct the vast majority of "problems" of this sort with their own formulae. As a professional, you probably already know this, but when you budget for your camera, lenses, flash, etc. Also include a margin for PhotoShop (expensive), QimagePro (VERY reasonable), Genuine Fractals (between
Lin
Robert MacLellan
Senior Member
I think you will find that you may NEVER get a "really good looking 16x20 or 20x24" out of a digital camera even with Genuine Phony...sorry... Genuine Fractals.
A 16x20 would require an image of 4800x6000 and a 20x24 would need a digital image of 6000x7200. Since the D1 is only going to give you 2012x1324 worth of information there isn't any 3rd party program that is going to be able to add those pixels that are missing and the best it can do is interpolate with phony pixels by guessing. Even a 3.3Mp camera will only deliver 2048x1536...far below what you are hoping for.
IF you want digital prints of that size...stick to film and a high resolution scanner...or... just use conventional optical enlargement technology which is much more satisfactory. On the other hand, the D1 should produce a truly excellent 4x6 or 5x7 but do you want to spend $5,000 for that?
A 16x20 would require an image of 4800x6000 and a 20x24 would need a digital image of 6000x7200. Since the D1 is only going to give you 2012x1324 worth of information there isn't any 3rd party program that is going to be able to add those pixels that are missing and the best it can do is interpolate with phony pixels by guessing. Even a 3.3Mp camera will only deliver 2048x1536...far below what you are hoping for.
IF you want digital prints of that size...stick to film and a high resolution scanner...or... just use conventional optical enlargement technology which is much more satisfactory. On the other hand, the D1 should produce a truly excellent 4x6 or 5x7 but do you want to spend $5,000 for that?
L
Lin Evans
Guest
Hi Robert,
Sounds like you don't like Genuine Fractals
Actually, we have had excellent results with it, up to and exceeding 16x20 with output from even 2 megapixel camers with the right subject material.I don't mean average results, I'm seriously saying "stunning results!" If you have used Genuine Fractals and been dissappointed, it could be that you are doing something wrong or that your original material was not up to being enlarged. These are problems which take time to work through and it does take a bit of experience with GF to understand getting the best results from it.
If you have actually looked at native 8x10 prints made from the D1, I don't think you would be terribly surprised to find that is quite possible to get much larger output with the help of fractal interpolation. Let me remind you that at the last superbowl, last play in the end-zone, there were four Sports Illustrated photographers who attempted to capture the last play. Two had bad angles, one was reloading film and the only one to get "the" shot was using a D1. S.I. was quite concerned since it was a digital image and only 2.66 mp, but after running it through Genuine Fractals, it was "as good" as anything else they had from any of their film cameras (their quote). It made a great two page spread and many converts were made that day.
Noted professional photographer Vincent Versace who photographs many of the leading Hollywood celebrities uses a D1 and has made incredible prints from it at a size of 8x10 (feet - not inches) using Genuine Fractals.
The interpolation isn't a "guess", and the program isn't "phony," it's based on a quite sophisticated algorithm which, when presented with a good base to work with, can accurately and systematically add pixels to create incredible results.
Whether or not this method is satisfactory for this photographer's needs is a question only he can answer, but in fairness to both the camera and the software, he should probably make that decision based on practical hands-on experience. Maybe renting a D1 and trying it would be helpful and in his best interest.
Lin
Edward Woods
Leading Member
Would agree with Lin that the D-1 is an excellent camera -especially for professional photographers - the only problem is that it has only 2.7 megapixal resolution.
If you don't need something immediately, I would suggest waiting for the rumored upgrade to the D-1. At $5,000, it would be a shame to buy one now and see the upgrade in a few months.
With regard to Genuine Fractals, I have tried using for it for several months, trying to get the kind of results that Lin claims to get. No luck. Can't even get to 16 x 20 inches much less 16 x 20 feet. The only image of that size that I have seen was at the Nikon booth at PMAI and it looked dreadful.
I suggest that you download the free demo from the Altamira website and try it on some of your existing images. I have had good success using Photoshop 5 and re-sampling the images in the 16 bit mode. In a direct comparison, Genuine Fractals was a little better, but (in my opinion) not enough better to justify spending the money on it.
The key here is to try it for yourself. Doesn't matter if it works for someone else - but doesn't work for you.
If you don't need something immediately, I would suggest waiting for the rumored upgrade to the D-1. At $5,000, it would be a shame to buy one now and see the upgrade in a few months.
With regard to Genuine Fractals, I have tried using for it for several months, trying to get the kind of results that Lin claims to get. No luck. Can't even get to 16 x 20 inches much less 16 x 20 feet. The only image of that size that I have seen was at the Nikon booth at PMAI and it looked dreadful.
I suggest that you download the free demo from the Altamira website and try it on some of your existing images. I have had good success using Photoshop 5 and re-sampling the images in the 16 bit mode. In a direct comparison, Genuine Fractals was a little better, but (in my opinion) not enough better to justify spending the money on it.
The key here is to try it for yourself. Doesn't matter if it works for someone else - but doesn't work for you.
L
Lin Evans
Guest
Good advice for Ed. But remember that the output from the D1, even though it's only a 2.62 mp image is MUCH cleaner than the images Ed is working with from his 3.34 megapixel Olympus C3030Z. This does make a significant difference. Also, if you use any digital camera output with Genuine Fractals, be sure to shoot in the uncompressed mode. Artifacts from jpg files are not well tolerated in GF and will not produce satisfactory enlargements in most cases. It's definitely something you will want to test for yourself before spending your money. My suggestion would be to have someone you know send you a full uncompressed image from their D1 of a subject type that you will be using the camera for. Print an 8x10 on a high quality inkjet (Epson 870 should do it) and see if the 8x10 meets your needs. Next download GF's demo and carefully follow their instructions. Take the image you save at 300dpi or 400dpi and at the size you wish to print to your closest processing site that has a Fuji or other suitable printer and have it printed. If the quality is satisfactory, you have a solution. If not, wait until native resolutions catch up a bit.
Lin
L
Lin Evans
Guest
Hi Ed,
You might want to read this too .....
http://www.dpreview.com/forums/read.asp?forum=1014&message=362080
Lin
F
Francis
Guest
Whether a digital photo resolution is enough or not for desired enlargement is much dependent on how much details the object contains. For some simple objects like product shot, you will find D1 picture could be enlarged to even 20x24 & be comparable to film shot. I've been asked by my boss to take large group people shot for my colleague using a 3.3 Mp digital camera, the photo is intended to be printed in A4 size, finally it failed even with interpolation.
The failure is just a simple mathematics, just find out how much pixels is available to describe one man's face & how many pixels is available to describe the each eye. Intepolation can make satisfactory detail to become better but never can't make a worse detail (actually nearly discontinued detail) to become any better.
Francis C.F.P.
L
Lin Evans
Guest
Another user of Genuine Fractals whose report is that it exceeded expectations.....
Lin
http://www.dpreview.com/forums/read.asp?forum=1014&message=362244
Freya22770
Active member
- Messages
- 60
- Reaction score
- 0
It depends on the type of photograph that you are taking. If it is a technical subject with a great deal of fine detail you will need a resolution of around 7 line pairs/mm in the final print ( this is the limit of human vision corresponding to 1 minute of arc) . This resolution is well within the capabilities of a good 1440 d.p.i printer. If the work is pictorial 2 line pairs/mm is adequate.
A picture say 2000 pixels wide with a 12mm ccd could be enlarged to around 6" wide to give a print with 7 line pairs mm and this would appear unconditionally sharp to the human eye. For a pictorial print the same file could be enlarged to around 20" wide and still look very good.
I have tried genuine fractals and for pictorial work it seems to be superb. For technical subjects it does not work because it can not recreate fine detail that the original camera did not capture.
I recently had to photograph a subject for a technical presentation. Thiis subject had fine detail that was 20 microns wide. I used an Olympus C2000 with a reversed tamron 90mm macro lens, this gave me an image on the ccd that was twice life size. On a small print it looked OK but when projected directly from the laptop through a digital projector the image was hopeless. I tried every type of interpolation available but in the end I shot on 35mm 100 iso transparency and scanned at 2700 d.p.i and the results were excellent . This image was 10.9 M pixels i.e. 4 times bigger than the D1.
I know that the DI is a lot better than the C2000 but I'd want to see some pretty convincing proof before parting with that much money and as my main imterest is pictorial I certainly wont buy another digicam untill somebody produces a body that will give me the equivalent to a true rectilinear 17mm (i.e. > 90 degree horizontal angle). As far as I can see the best the D1 would do is 21mm with a 14mm prime lens
So its horses for courses. Digital is great for many subjects but it ain't there yet if you want serious resolution.
If you know that your customers will never want bigger than say 12x10 I would say go digital. If they might want bigger ,shoot on digital but take back up shots on 35mm or medium format and just pass the cost of processing and film on to the customer, thats what I do. bearing in mind that there are still some doubts as to the durability of inkjet prints you will porobably have to have laser scanned prints on conventional paper done and unless the negs/trannies are scanned at the time of processing you will probably find that digital prints from your own CD's will cost you more than the same prints straight from negs. They certainly do here in the UK.
Freya
A picture say 2000 pixels wide with a 12mm ccd could be enlarged to around 6" wide to give a print with 7 line pairs mm and this would appear unconditionally sharp to the human eye. For a pictorial print the same file could be enlarged to around 20" wide and still look very good.
I have tried genuine fractals and for pictorial work it seems to be superb. For technical subjects it does not work because it can not recreate fine detail that the original camera did not capture.
I recently had to photograph a subject for a technical presentation. Thiis subject had fine detail that was 20 microns wide. I used an Olympus C2000 with a reversed tamron 90mm macro lens, this gave me an image on the ccd that was twice life size. On a small print it looked OK but when projected directly from the laptop through a digital projector the image was hopeless. I tried every type of interpolation available but in the end I shot on 35mm 100 iso transparency and scanned at 2700 d.p.i and the results were excellent . This image was 10.9 M pixels i.e.
I know that the DI is a lot better than the C2000 but I'd want to see some pretty convincing proof before parting with that much money and as my main imterest is pictorial I certainly wont buy another digicam untill somebody produces a body that will give me the equivalent to a true rectilinear 17mm (i.e. > 90 degree horizontal angle). As far as I can see the best the D1 would do is 21mm with a 14mm prime lens
So its horses for courses. Digital is great for many subjects but it ain't there yet if you want serious resolution.
If you know that your customers will never want bigger than say 12x10 I would say go digital. If they might want bigger ,shoot on digital but take back up shots on 35mm or medium format and just pass the cost of processing and film on to the customer, thats what I do. bearing in mind that there are still some doubts as to the durability of inkjet prints you will porobably have to have laser scanned prints on conventional paper done and unless the negs/trannies are scanned at the time of processing you will probably find that digital prints from your own CD's will cost you more than the same prints straight from negs. They certainly do here in the UK.
Freya
Robert MacLellan
Senior Member
Well, I remember the howl that went up when Fuji had the affrontery to offer interpolation in their camera. Everyone screamed in indignation and the anti-Fuji posts were flying furiously around the web. Any camera that uses the "interpolation" is regularly put down as somehow..sub-standard.
At the same time Genuine Fractals gave other users the opportunity to secretly apply interpolation themselves without anyone being the wiser and...not only was there NOT any criticism of this deception but it is being actively promoted. Sounds a lot like a double standard to me! It is not that I dislike Genuine Fractals...it is just a program... but hypocricy gets my attention.
I am familiar with Mr Versace's work and his frequent appearances in photo magazines...the less said, the better.
I am sure you realize that there are no algorithms, no matter how sophisitcated they may be, that can recreate reality that has not been recorded. It may look good after the fact when you don't have reality to compare it with but, there is no doubt that it is an invention...a calculation...a clever trick..but that is all.
At the same time Genuine Fractals gave other users the opportunity to secretly apply interpolation themselves without anyone being the wiser and...not only was there NOT any criticism of this deception but it is being actively promoted. Sounds a lot like a double standard to me! It is not that I dislike Genuine Fractals...it is just a program... but hypocricy gets my attention.
I am familiar with Mr Versace's work and his frequent appearances in photo magazines...the less said, the better.
I am sure you realize that there are no algorithms, no matter how sophisitcated they may be, that can recreate reality that has not been recorded. It may look good after the fact when you don't have reality to compare it with but, there is no doubt that it is an invention...a calculation...a clever trick..but that is all.
Hi Robert,
Sounds like you don't like Genuine Fractals
excellent results with it, up to and exceeding 16x20 with output from
even 2 megapixel camers with the right subject material.
I don't mean average results, I'm seriously saying "stunning results!" If
you have used Genuine Fractals and been dissappointed, it could be that
you are doing something wrong or that your original material was not up
to being enlarged. These are problems which take time to work through and
it does take a bit of experience with GF to understand getting the best
results from it.
If you have actually looked at native 8x10 prints made from the D1, I
don't think you would be terribly surprised to find that is quite
possible to get much larger output with the help of fractal
interpolation. Let me remind you that at the last superbowl, last play in
the end-zone, there were four Sports Illustrated photographers who
attempted to capture the last play. Two had bad angles, one was reloading
film and the only one to get "the" shot was using a D1. S.I. was quite
concerned since it was a digital image and only 2.66 mp, but after
running it through Genuine Fractals, it was "as good" as anything else
they had from any of their film cameras (their quote). It made a great
two page spread and many converts were made that day.
Noted professional photographer Vincent Versace who photographs many of
the leading Hollywood celebrities uses a D1 and has made incredible
prints from it at a size of 8x10 (feet - not inches) using Genuine
Fractals.
The interpolation isn't a "guess", and the program isn't "phony," it's
based on a quite sophisticated algorithm which, when presented with a
good base to work with, can accurately and systematically add pixels to
create incredible results.
Whether or not this method is satisfactory for this photographer's needs
is a question only he can answer, but in fairness to both the camera and
the software, he should probably make that decision based on practical
hands-on experience. Maybe renting a D1 and trying it would be helpful
and in his best interest.
Lin
L
Lin Evans
Guest
Fuji certainly wasn't the first to put interpolation in their cameras - as I remember Kodak had a model or so with it and Epson is doing it as we speak, and the "howl" wasn't that Fuji had interpolation, it was that they were not up-front about it (I seem to remember that the high resolution 4.whatever sticker was removed from their FinePix cameras). It was the perceived false advertising which raised hackles, not the fact that Fuji was using what they lovingly call "extrapolation".
Where you see hypocracy is beyond me? If you are truly familiar with Vincent Versace's work, then you are undoubtedly aware that he is recognized as one of the leading professional photographers in the business and makes more money every month than you and I together will probably make in the next 10 years. He must be doing something which endears himself to his clients!
You are absolutely correct in saying that no algorithm can create reality which hasn't been recorded. That's the first thing I point out to anyone intending to use GF. Use it for photos which are amenable to it - not all are, and I carefully explain the caveats and what to expect.
Yes, GF is an invention. The camera is an invention. What's your point? It's not a "clever trick", its a tool just like your camera. When used with the proper data it can produce incredible results. The thousands of satisfied professional users can attest to that fact. When fed garbage in, it returns the same. Lots of us love it for what it can help us do with our marginal equipment. If you don't like it, or don't understand how to use it, then don't. No one I know is "secretly" applying interpolation, "without anyone being the wiser". If you think it's that good, then you just shot down your own argument against it.
If you don't like it, just say you don't like it and be done with it - but let the user who has yet to experience it decide for themselves. I can only report my own first hand experience with it and I find that it works very well for me. Many others are reporting the same results. Perhaps your standards are higher than those who find it to be useful, but if they were, I'm wondering why you are using digital at all???
Lin
About fifteen years ago I'd practiced professional photography for a number of years and used everything from 4x5 view cameras with swings and tilts as well as medium formats and 35 mm. Now's it's a hobby. I know very little about digicams. But what counts and what is always paid for is the excellence of the final product (the photo) resulting from the skills of the photographer. All this sounds like the old argument of can a good 35mm or even a good medium format camera match a good 4x5 view camera with its swings and tilts? Just for fun let's say can it match an even bigger view camera for product photography? The answer is certainly not. There is convenience and speed in handling smaller formats but that's all, and that's a lot for grab images, or after digicam perfection even fashion photography. Each tool should be used as intended. The fine results I've seen and examined after downloading are impressive for what they are. For what they are, because film scans of my old negatives have by far better contour sharpness. Perhaps in another ten years will put the majority of conventional cameras out of business.
Rick Maiman
Well-known member
Thank you Lin, for the most cogent and vividly sensable explanation of just what "Fractals" does, and, is capable of doing.
Thanks, Rick Maiman
Thanks, Rick Maiman
Frances, I'm assuming you shot this with the 990. Did you try Genuine Fractals with the shot?
mike.
I don't think this is the case, Robert. A camera that billed itself as a 2Mp camera with a 4.7MP interpoloation feature is fine. In fact, It should stand out among the rest of the 2MP cameras for that feature. To market it as a 4MP camera, however, is deceptive and the amnufacturer deserves a stinging rebuke for the ploy.
I disagree that this is the case at all, on a number of points. 1) GF didn't claim to be selling something that they weren't (eg, a 4MP camera). 2) There is an emormous difference in the algorithms being used that is reflected in the quality of the final product. 3) There are excellent reasons to dislike image expansion (interpolation or fractal) on a camera, but like it on the desktop. Flash Memory real estate is at a premium in these multi-MP cameras. It makes little sense to expand the image on camera when the expansion could be done on computer and the effective capacity of the flash memory thereby doubled. 4) There were (in the 4700's case) good reasons to believe that the interploation algorithm used by the camera was inferior to other algorithms available in software.
I don't think that comment is called for. Perhaps you could point us to some of your more recent publications so we could learn from the comparison.
Here we are in agreement. It is a clever trick -- the make the eye believe that there is detail in the print that the camera didn't capture. I doubt that there are many of us that wouldn't prefer a 12MP, 24MP, or 96MP camera if they were technically and economically viable. Until then clever technology like Genuine Fractals offers a reasonable means for many photographers to get the most our of there 2 and 3MP cameras. It's a clever trick appauded by those who know how to use it.
Edward Woods
Leading Member
Lin:
I checked your link and he says that he used Genuine Fractals and that
the results were "pretty good". Afraid that my expectations are a lot
higher than that. Also the images he posted is only screen resolution
so there is no way to tell if GF made any improvements in the image.
Ed
I checked your link and he says that he used Genuine Fractals and that
the results were "pretty good". Afraid that my expectations are a lot
higher than that. Also the images he posted is only screen resolution
so there is no way to tell if GF made any improvements in the image.
Ed
Edward Woods
Leading Member
Lin:
I checked your link and he claims to have gotten good
results - it's just conversation and he claims to have
an expert who says you can't get more than an 8 x 10 from
a Fuji S-l. I know that is wrong so who do you believe.
From my experience, if the original image is of high quality
than you can use Photoshop (in 16 bit mode) to increase the resolution
and print larger size prints. There is a limit to this since
interpolation results in artificial not true resolution.
Lin has taken some great shots with the D-l and uses the
raw format. I have been told that if you have this type of
format that Genuine Fractals works much better than if you
try it on a JPEG for example. Unfortunately most digicams
do not offer a raw format. We all wish that they did.
All I am trying to say is that potential customers for GF
should download the trial version from the Altamira website
and use it to upsize their images and see if the program works
up to their personal expectations.
Ed
I checked your link and he claims to have gotten good
results - it's just conversation and he claims to have
an expert who says you can't get more than an 8 x 10 from
a Fuji S-l. I know that is wrong so who do you believe.
From my experience, if the original image is of high quality
than you can use Photoshop (in 16 bit mode) to increase the resolution
and print larger size prints. There is a limit to this since
interpolation results in artificial not true resolution.
Lin has taken some great shots with the D-l and uses the
raw format. I have been told that if you have this type of
format that Genuine Fractals works much better than if you
try it on a JPEG for example. Unfortunately most digicams
do not offer a raw format. We all wish that they did.
All I am trying to say is that potential customers for GF
should download the trial version from the Altamira website
and use it to upsize their images and see if the program works
up to their personal expectations.
Ed
lloyd chambers17667
Active member
- Messages
- 62
- Reaction score
- 0
I think it misses the point to talk only about resolution; the human brain forms a visual impression based on a lot more than resolution.
There are other factors which make a picture look attractive, including dynamic range, color saturation and amount of noise or grain. The D1 excels in these areas. I have been extremely pleased with 13 X 19 prints made from the D1. It is true that the absolute amount of detail is not there as compared to say, Velvia, but the visual impact is stunning due to other factors. Plus you can can easily enhance a digital file as a first generation file, rather than a scan from film.
Lloyd
There are other factors which make a picture look attractive, including dynamic range, color saturation and amount of noise or grain. The D1 excels in these areas. I have been extremely pleased with 13 X 19 prints made from the D1. It is true that the absolute amount of detail is not there as compared to say, Velvia, but the visual impact is stunning due to other factors. Plus you can can easily enhance a digital file as a first generation file, rather than a scan from film.
Lloyd
I'm personally not a fan of GF. Not because it does not work - I just think the price tag is too high for what you get.
On a variety of resized examples, the Lanczos interpolation provided by Q-Image Pro is more than acceptable compared to GF and Photoshop's bi-qubic. However, there are a large number of cases where GF is slightly better than the bi-cubic and/or Lanczos interpolation. Of course, IMO the points of contention are the word "slightly", and the high cost of GF.
Just for talking purposes, there is an example at the following link where IMHO, the GF interpolation is "slightly" better than Lanczos or bi-cubic. Go for the "Interpolation" section.
http://members3.clubphoto.com/joe243826/
I think the example is representative of Lin's thoughts regarding knowing when you should (or should not) use a particular tool for your application. If I made my living in the photographic field, GF would be worth the money (i.e. it would be available when it becomes the right tool). I have compared the three algorithms 13 or 14 times on different images - it does pay to know when you will have a successful GF resize depending on the image content.
Because of my initial choice of images (low contrast, shadow detail), there was no clear winner the first two times I compared the algorithms. I was very confused about my initial test results. We all have individual opinions, and I don't always agree with Lin. However, I consider Lin to be a VERY sharp guy and it bothers me when my results don't agree with individuals that have gained my respect. Eventually I found a "family" of images where GF was clearly better (again, IMHO).
I did my thing and am satisified that GF is not a tool that I need. However, I'll bet $1,000,000 that it "could" be an important tool for you. Everyone should try it on at least ten different types of images that are normally recorded. If GF improves the image for your application even 30% of the time, it falls into the category of an "important" tool.
Hope this real world example helps quantify the discussion for your 16x20/20x24 applications. Also, FWIW, the D1 is a superb camera from the standpoint of "low noise" in shadows. I don't own a D1 - simply ran a few tests with a D1 to see what is going on compared to my one and only digital camera and scanned images on film. If your main need is 8x10, I think the D1 is a "killer" tool (provided you correct for the horrible color). All (six) of the 16x20 prints I've seen from D1 files have looked very good to me; all prints were portraits.
Sincerely,
Joe Kurkjian
On a variety of resized examples, the Lanczos interpolation provided by Q-Image Pro is more than acceptable compared to GF and Photoshop's bi-qubic. However, there are a large number of cases where GF is slightly better than the bi-cubic and/or Lanczos interpolation. Of course, IMO the points of contention are the word "slightly", and the high cost of GF.
Just for talking purposes, there is an example at the following link where IMHO, the GF interpolation is "slightly" better than Lanczos or bi-cubic. Go for the "Interpolation" section.
http://members3.clubphoto.com/joe243826/
I think the example is representative of Lin's thoughts regarding knowing when you should (or should not) use a particular tool for your application. If I made my living in the photographic field, GF would be worth the money (i.e. it would be available when it becomes the right tool). I have compared the three algorithms 13 or 14 times on different images - it does pay to know when you will have a successful GF resize depending on the image content.
Because of my initial choice of images (low contrast, shadow detail), there was no clear winner the first two times I compared the algorithms. I was very confused about my initial test results. We all have individual opinions, and I don't always agree with Lin. However, I consider Lin to be a VERY sharp guy and it bothers me when my results don't agree with individuals that have gained my respect. Eventually I found a "family" of images where GF was clearly better (again, IMHO).
I did my thing and am satisified that GF is not a tool that I need. However, I'll bet $1,000,000 that it "could" be an important tool for you. Everyone should try it on at least ten different types of images that are normally recorded. If GF improves the image for your application even 30% of the time, it falls into the category of an "important" tool.
Hope this real world example helps quantify the discussion for your 16x20/20x24 applications. Also, FWIW, the D1 is a superb camera from the standpoint of "low noise" in shadows. I don't own a D1 - simply ran a few tests with a D1 to see what is going on compared to my one and only digital camera and scanned images on film. If your main need is 8x10, I think the D1 is a "killer" tool (provided you correct for the horrible color). All (six) of the 16x20 prints I've seen from D1 files have looked very good to me; all prints were portraits.
Sincerely,
Joe Kurkjian
Keyboard shortcuts
- f
- Forum