Re: Enlarging digital images - examples
Mike Warren
> This, it seems got up the nose of those who demand to see the magic
> pixels, instead of judge the photographs made after using my process,
> they want to become "pixel peepers".
About me, you are wrong there. I want to see a sharper versions
of my pictures when enlarged. You claim to be able to do that
better than anyone else.
I want to see one of *my* photos enlarged by your process. How
do I get that?
-Mike
Paul Furman
> http://www.canvasphotos.com.au/enlarge-example2.htm
>
> ... I elected to
> take the line of least resistance and publish visual size examples
> without even attempting to do the impossible. After all when you see a
> photograph it is the enlarged size, why not make the monitor picture
> that size too?
As others said we don't doubt you can make nice enlargements, but I'm
still skeptical about these demonstrations. It would be the simplest
thing to crop the final result as you appear to have done here but why
is the white balance different on the crop?
I tried shrinking that crop down to native pixels and I don't believe
you could have got the detail in the eye if those were the native pixels:
http://www.edgehill.net/1/?SC=go.php&DIR=Misc/misc-photos/temp
-my sizing is approximate but I don't think it is anywhere near the
possibility of doing this. I figure the native camera pixels of your 506
pixel enlarged crop at about 127 pixels, that's a 4x enlargement.
Scott W
> Colin D <Col...@killspam.127.0.0.1> wrote in
> news:4434F9D4...@killspam.127.0.0.1:
>
>
> >
> > Giving Doug the benefit of the doubt here, maybe he wasn't fiddling
> > his images for this thread, but, it's still on a monitor. A print is
> > the definitive proof, and we don't have a print. Some of you Oz types
> > should be able to get to Brizzie and check out a real print. It's a
> > bit far from here (New Zealand) {:-)
> >
> > Colin D.
> >
>
>
> For the lone person (a true -retired- Professional Photographer, at
> that) who demonstrates at least the ability to comprehend what I have
> been saying and doing for many years, might actually be possible...
>
> Send your postal address to: d-mac AT canvasphotos DOT com DOT au and
> I'll respond with a beautiful Giclée print of this picture at 24"x36"
> and you can then claim to be one of few people posting in these groups
> who actually has a genuine "techno Aussie Giclée enlargement" and can
> speak with the authority of someone with first hand knowledge.
> http://www.canvasphotos.com.au/enlarge-example2.htm I'll send it at no
> charge to you, Colin.
>
> Because I recognize the stupidity of trying to display a 350 megabyte
> photograph on a computer monitor which may or may not be able to display
> anywhere near the correct colour, much less be in focus... I elected to
> take the line of least resistance and publish visual size examples
> without even attempting to do the impossible. After all when you see a
> photograph it is the enlarged size, why not make the monitor picture
> that size too?
>
> pixels, instead of judge the photographs made after using my process,
> they want to become "pixel peepers".
>
> judgement who did get enlarged photographs, all made positive comments.
> Gordon Moat wrote an article about the process, saying he at first
> didn't believe it was possible. He does now.
> http://www.allgstudio.com/technology.html
>
> Maybe when you get yours, Colin, you can tell us your opinion?
First off a few words about your idea that anyone who make a copy of
your image is stealing from you. Consider the "Selling it" section of
Consumer Reports, this is the part where they show the dumb and
misleading things companies say when promoting their produces. As part
of this they often reproduce ads and make fun at them. These ads are
of course copyrighted and in most cases the last place you want your ad
to show up is in the selling section of Consumer Reports. So what makes
you think your claims are not subject to the same kind of editorial
scrutiny that everyone else goes through?
Now onto your image, first it is a bit deceiving since your indicated
crop area in the full image is much smaller then the actual crop. The
crop looks pretty soft to me and something that I would expect any good
interpolation program to be able to produce, but then you did not
provide the original image at full resolution so that other might see
how your method stacks up against others methods.
Lastly I am always a bit distrustful when the person doing the are
using their own image.
If you really do care about convincing people that you have something
that is better then other methods then you would use a test image from
an independent source.
In the end we all pretty much believe you can make good looking big
prints, the part that non of us take on faith is that you have ways of
doing this that are better then current algorithms.
You of course are under no obligation to prove anything to anyone, but
you seem determined to do so. If you really want to prove your methods
it would be very easy to do so.
Scott
Somebody
> judgement who did get enlarged photographs, all made positive comments.
> Gordon Moat wrote an article about the process, saying he at first
> didn't believe it was possible. He does now.
> http://www.allgstudio.com/technology.html
Exactly where does Gordon Moat say that? Immediately after saying he'd
received your images, without commenting on anything except that you'd
"pushed the files from 8 MP to around 50 MP ", he says:
"One thing that needs to be considered is that upsizing any image will never
add information to the original file. What first happened in the camera
becomes the limit of detail, regardless of post processing, or algorithms."
That's hardly any support for your claims, Douglas!
Perhaps Gordon Moat isn't so qualified after all. Oh, BTW, what makes
someone qualified to say whether an upsized image of yours actually has more
detail than the original?
Gordon Moat
Somebody wrote:
>>What really puzzles me most of all is that the people qualified to pass
>>judgement who did get enlarged photographs, all made positive comments.
>>Gordon Moat wrote an article about the process, saying he at first
>>didn't believe it was possible. He does now.
>>http://www.allgstudio.com/technology.html
>
>
> Exactly where does Gordon Moat say that? Immediately after saying he'd
> received your images, without commenting on anything except that you'd
> "pushed the files from 8 MP to around 50 MP ", he says:
> "One thing that needs to be considered is that upsizing any image will never
> add information to the original file. What first happened in the camera
> becomes the limit of detail, regardless of post processing, or algorithms."
>
> That's hardly any support for your claims, Douglas!
>
First off, thanks for reading my article. I should point out that the
original requests for me to view images samples did not come directly
from Douglas. I should also point out that the original claim that
Douglas had was that his algorithm did not lose any detail information.
I am not sure how or when that changed in the last year to a claim of
adding or increasing detail information; it is simply not possible
regardless of technology.
> Perhaps Gordon Moat isn't so qualified after all. Oh, BTW, what makes
> someone qualified to say whether an upsized image of yours actually has more
> detail than the original?
>
>
Okay, so to make this really, really simple:
Question: Can Douglas make nice large prints?
Answer: Yes.
Q: Does the algorithm Douglas developed work as well as other
commercially developed interpolation algorithms?
A: Yes.
Q: Can someone with a reasonable level of professional experience,
knowledge, and commercially available algorithms, and using a
professional level wide format inkjet system, reproduce the results
Douglas obtained?
A: Yes.
Q: Why did I write those articles?
A: I was working on a proposal for funding to include wide format
imaging and high end scanning solutions. This would be to supplement my
existing photography and graphic design for print business. In other
words, I did not write the article for Douglas, and I did not write
those articles for the pleasure of newsgroup people. One nice aspect
that happened as a result of those articles was being subcontracted to
help develop a future full frame digital SLR, which was an interesting
project that happened in May last year. The other nice aspect of those
articles is they support some technical aspects of my funding proposal.
Q: Did Douglas's images contain more detail information than the original?
A: No, only more pixels.
Q: What does "more pixels" mean?
A: The additional pixels are interpolated data, likely due to one or
several algorithms.
Q: What qualifications do I have?
A: Experience in surveillance, commercial printing, development of
colour models for imaging software (consultant or collaborative team
environment), and commercial photography. I also have used numerous
software applications for imaging, including PhotoShop since version
2.5, nearly every day since 1994. I don't doubt that someone might be
more qualified, or have different qualifications.
I was asked for comments, I happened to be writing an article and doing
research at the time, so I don't mind sharing my comments. That leaves
people free to explore more, do their own research, and draw their own
conclusions; since you read the article, you probably notice I
encouraged that.
Ciao!
Gordon Moat
A G Studio
<http://www.allgstudio.com>
Scott W
D-Mac wrote:
> them and me.
> I have never claimed to be able to add picture elements like my
troll has
> created the impression adding detail should do. I can't add people who are
> not in the original or lines of text that were never there either. I do add
> detail in the form of pixels of predictable colour so that the picture
> retains its definition and clarity during enlargement.
I think it is statments light this that get people questioning you
"even adding detail which was never there in the first place. "
For those who want to see the whole thread here is the link
http://groups.google.com/group/comp.graphics.apps.photoshop/browse_frm/thread/90dff1d6c6425889/
And you also made this statment
"this process of mine not just works but works better than any
commercially available software running under Windows or Apple
operating
systems." but have never backed it up with anything but words.
Scott
Mr.T
"D-Mac" <no....@nt.use.1> wrote in message
news:Xns97A0AB9496...@61.9.191.5...
> About 2 years ago, I wrote that I had to add detail to the image to
> maintain its sharpness. This is correct. I make the same claim today. If I
> didn't, the image would simply not enlarge. The detail I add is in the
form
> of pixels my algorithm predicts as being most appropriate for the job.
>
> I have never claimed to be able to add picture elements like my troll has
> created the impression adding detail should do. I can't add people who are
> not in the original or lines of text that were never there either. I do
add
> detail in the form of pixels of predictable colour so that the picture
> retains its definition and clarity during enlargement.
Just like any interpolation program does then. You would need to show it is
better than Genuine Fractals for instance?
Even that is hard to pick over using stepped interpolation in Photoshop.
> My process sometimes creates vectorised edges of all clearly definable
> edges when I get a soft image to enlarge too. I overlay this "edge map" on
> the finished enlargement that creates the impression of an image much
> sharper than it actually is.
Like using Photoshops "unsharp mask" then?
> So just in case you are not a troll... I add detail in the form of
> additional pixels to make an enlargement from a digital image with no
> visible loss of detail. This is all I have ever claimed, regardless of
what
> the troll would make out.
Fair enough too, if you aren't claiming superiority for your algorithm?
MrT.
Bart van der Wolf
"Scott W" <bip...@hotmail.com> wrote in message
news:1144518371.7...@j33g2000cwa.googlegroups.com...
SNIP
> If you really do care about convincing people that you have
> something that is better then other methods then you would
> use a test image from an independent source.
My thoughts exactly.
> In the end we all pretty much believe you can make good
> looking big prints, the part that non of us take on faith is that
> you have ways of doing this that are better then current
> algorithms.
Again, I 100% agree. A *comparison* between output from different
methods is the only valid test.
> You of course are under no obligation to prove anything to
> anyone, but you seem determined to do so. If you really want
> to prove your methods it would be very easy to do so.
And such an easy test has been offered:
<http://heim.ifi.uio.no/~gisle/photo/interpolation.html>
Although I could understand that the part that gets a bit personal
might reduce Douglas' enthousiasm to cooperate.
Bart
POTD.com.au
"D-Mac" <no....@nt.use.1> wrote in message
news:Xns97A0AB9496...@61.9.191.5...
> news:49r0u4F...@individual.net:
>
>>
>> Perhaps Gordon Moat isn't so qualified after all. Oh, BTW, what makes
>> someone qualified to say whether an upsized image of yours actually
>> has more detail than the original?
>>
>>
>>
>>
>
> become accountable to them is reason enough to be wary of the question.
> The
> nature of the "detail" part of your question also makes me think you never
> read the original statement on this subject but instead, use the version
> created by my troll and his followers.
>
> I'll give you the benefit of the doubt and answer your question anyway.
>
> I provided 2 sample enlargements to Gordon and the original Camera RAW
> files. He acted as an independent - qualified as in legally addressable
> qualifications - arbitrator to keep someone else from claiming my prints
> from a 10D and 20D were not equal to Medium format enlargements.
>
> Gordon was under no obligation to do or say anything other than tell the
> truth as he found it, should his expertise be required if the other party
> decided to play dirty. His article was something I had nothing to do with
> other than my pictures giving him the idea to write it.
>
> The other party didn't attempt to claim anything after seeing my prints.
> His reply started out with: "...The prints arrived in all their glorious
> colour and detail" His later comments which some may construe to be less
> flattering were related to the lpi of the Grand Sherpa Printer I used
> rather than any contradiction of my claims.
>
> There has been a shift in this topic over the past 2 years from people
> claiming I am simply not able to make these large prints to people like
> Mike Warren who are looking now not for proof I can make a clear and sharp
> enlargement but for a reason, any reason they can use to discredit my
> claim
> of adding detail to facilitate an enlargement. The most obvious
> contradiction here, is the different interpretation of "detail" between
> them and me.
>
> You can't enlarge a digital image without adding pixels derived from a
> mathematical prediction of what colour they need to be. Any ADDED pixel
> different from a 255 255 255 colour is detail. If you tried to enlarge an
> image using pure white pixels, all you'd get is a picture with jagged
> edges
> in between a large amount of white space... NO DETAIL.
>
> About 2 years ago, I wrote that I had to add detail to the image to
> maintain its sharpness. This is correct. I make the same claim today. If I
> didn't, the image would simply not enlarge. The detail I add is in the
> form
> of pixels my algorithm predicts as being most appropriate for the job.
>
> I have never claimed to be able to add picture elements like my troll has
> created the impression adding detail should do. I can't add people who are
> not in the original or lines of text that were never there either. I do
> add
> detail in the form of pixels of predictable colour so that the picture
> retains its definition and clarity during enlargement.
>
> edges when I get a soft image to enlarge too. I overlay this "edge map" on
> the finished enlargement that creates the impression of an image much
> sharper than it actually is.
>
> additional pixels to make an enlargement from a digital image with no
> visible loss of detail. This is all I have ever claimed, regardless of
> what
> the troll would make out.
>
> mine.
>
Not true!
See thread "I was in the Redland News office today"
mark.t...@gmail.com
>His (Gordon Moat's) article was something I had nothing to do with
>other than my pictures giving him the idea to write it.
Anyone who still takes douglas at his 'word' should read the following
response from Gordon Moat.
http://groups.google.com/group/rec.photo.equipment.35mm/browse_frm/thread/f0bc2af22e84217b/
You'll note that Gordon gently clarifies his position in a way that..
er.. seems to contradict Douglas' version rather signficantly. But
read it for yourself.
mark.t...@gmail.com
> I am not sure how or when that changed in the last year to a claim
>of adding or increasing detail information; it is simply not possible
>regardless of technology.
Agreed. There are of course some limited opportunities offered by
deconvolution and richardson-lucy-ish stuff, but that isn't what is
being claimed by douglas. Those claims about more 'real' detail came
about because of this thread, amongst many others:
http://groups.google.com/group/comp.graphics.apps.photoshop/browse_frm/thread/90dff1d6c6425889
The title of the thread was: "Enlarged digital images with more detail
than the original" and it was of course, penned by douglas.
In that thread, he states:
"..many (of the enlargements) have more detail ... than the originals."
If that's not enough he goes on to include the following wonderful
paragraph:
=== Douglas ===
Pictorial evidence that it is indeed possible to enlarge a digital
image
which has a normal print size of 6.5" x 10" at 300 dpi, to 24" x 36"
poster
print with 720 dpi and still maintain the same sharpness and detail -
even
adding detail which was never there in the first place.
=== end ===
In a related thread of around the same date, he also states:
=== Douglas ===
Someone else claimed I couldn't enlarge a photo and at the same time
add
enough detail the image so a number plate which was unclear, could be
read.
All this demonstration does is to offer visual examples with print size
crops, that I most certainly can do this.
=== end ===
Needless to say the image he posted showed *no such thing*, and of
course, has now been withdrawn.
Gordon Moat
mark.t...@gmail.com wrote:
> Thanks for that very enlightening post, Gordon!
>
>>I am not sure how or when that changed in the last year to a claim
>>of adding or increasing detail information; it is simply not possible
>>regardless of technology.
>
>
> Agreed. There are of course some limited opportunities offered by
> deconvolution and richardson-lucy-ish stuff, but that isn't what is
> being claimed by douglas. Those claims about more 'real' detail came
> about because of this thread, amongst many others:
>
> http://groups.google.com/group/comp.graphics.apps.photoshop/browse_frm/thread/90dff1d6c6425889
>
> The title of the thread was: "Enlarged digital images with more detail
> than the original" and it was of course, penned by douglas.
>
> In that thread, he states:
> "..many (of the enlargements) have more detail ... than the originals."
>
Sorry, I don't have much interest in following the PhotoShop groups. I
see enough PhotoShop questions on PDN.
One needs to be careful about using the word "detail". Just an example:
double the number of pixels in an image, and there will be more pixels;
this is a simple process. If a pixel can define a detail, then we have
added detail. What then needs to be considered is what exactly is that
detail. On a simple level, it is likely only colour information for
those added pixels. Depending upon software, the added pixels will
retain a colour value somewhere between the range of the adjacent
pre-existing pixels. This is how most interpolation procedures function,
at least on a very simple level. Anyone who wants to get really
technical on this should just talk to Thomas Knoll.
Okay, so what photographers often term as "detail" is distinct elements
of information, as opposed to distinct pixels. This is what I refer to
in my statement. The original "detail" was contained in the capture,
whether film or direct digital. Depending upon how one defines "detail",
the interpretation of statements could be different.
> If that's not enough he goes on to include the following wonderful
> paragraph:
>
> === Douglas ===
> Pictorial evidence that it is indeed possible to enlarge a digital
> image
> which has a normal print size of 6.5" x 10" at 300 dpi, to 24" x 36"
> poster
> print with 720 dpi and still maintain the same sharpness and detail -
> even
> adding detail which was never there in the first place.
> === end ===
>
If Douglas defines "detail" as pixels of colour information, then he is
correct. However, I do not consider that proper usage.
> In a related thread of around the same date, he also states:
>
> === Douglas ===
> Someone else claimed I couldn't enlarge a photo and at the same time
> add
> enough detail the image so a number plate which was unclear, could be
> read.
> All this demonstration does is to offer visual examples with print size
> crops, that I most certainly can do this.
> === end ===
>
Sharpening and masking. Probably lots of people had already done this.
In this statement it seems that apparent sharpness of the end print has
been used to define "detail". Changing something from fuzzy and unclear
to readable is not adding "detail"; there is still nothing beyond the
original capture information.
In the flower print Douglas sent me, I saw evidence of sharpening and
masking. To his credit, there was almost no indication of halos nor
aberrations due to sharpening and masking. The only exception was in the
shadow areas, though your really had to look to find it, and I needed a
loupe to confirm it.
An independently developed interpolation algorithm with sharpening and
masking procedures could be of interest to companies making imaging or
printing products. There are existing products on the market, and I
think using one, or a combination of a few, could give someone nearly
the same results. The difference is that an individual might have little
interest in what Douglas has done, beyond possibly having him do some
nice large prints. The uniqueness, if any, would be of more interest to
a corporation that wants an alternative to licensing existing products
that could provide similar results.
Douglas can lose his temper and make outbursts and claims that make
others question everything else he states. Maybe I am a little
idealistic, but I like to focus on the good things people are capable of
achieving; despite his temper and outbursts, or even wild claims, he can
make nice large prints.