Do Sony users know this about photoshop?

I recently started resizing and printing from Photoshop 7. Previously I used CorelDraw 8. While I found that Photoshop printed images with better color density, I was shocked to learn that when you resize in photoshop (which usually means downsampling for dslr users), you loose pixel data and as a result you loose detail and sharpness. Fortunately there's a workaround and that's the subject of this post.

I stumbled on this information while reading "Real World PHotoshop 7". The workaround is to convert your image to a vector file before downsampling (either in the later versions of photoshop) or in a draw program such as Adobe Illustrator or CorelDraw, then resize. You can then post process and sharpen in Photoshop and print, but do not change the size there. Of course, PHotoshop is not the only digital (or raster) paint program that requires this work around.

I tested this by opening and resizing an image in CorelDraw (a raster/draw program). Then I exported the image as a tiff file to my documents. Next I opened the original version of the same image in photoshop 7 and resized it to the same size as the image from CorelDraw. After I opened the CorelDraw tiff version and aligned it next to the PHotoshop version, I zoomed onto each image to compare detail and sharpness. The differences were dramatic. The CorelDraw (vector) version retained all the detail and sharpness of the original, while the Photoshop (raster/bitmap) version did not. The differences are very visible when printing, even after sharpening. The differences are less apparent when you enlarge a scanned image in the respective programs. But dslr users are almost always downsizing, even when printing posters.

I entered this post here in Sony because I'm constantly reading where users resize in photoshop. Probably most of you already have this information. But since we downsize and compress images for viewing on this web site and since image detail and sharpness is such an important dynamic in dslr photography, I thought for those who do not know, this could be an important insight.

CHaCHa

CHa CHa:

Have you tried this starting with an image from your camera? Or did you first create a vector graphic in CorelDraw and work from that?

AFAIK it is impossible to start with a bitmap graphic and convert it to a vector graphic that retains all the detail and nuances of the original. Vector graphics by their nature are made up of mathematically-defined shapes and fills which are resolution-independent.

If you start with a vector graphic and then apply the steps you described, of course the downsized vector graphic will look better than the downsized bitmap. And as you zoom in to the vector graphic, the program will always render it at full screen resolution, so it will appear to retain more detail and sharpness.

So I don't think the method you are describing here can be applied to images that start out as bitmaps.

-- Bill
http://billw.smugmug.com

The difference you see is due to the difference between Rastor and Vector. Vector drawings are mathematical representations of all the shapes in the drawing. When you resize a vector drawing the image is redrawn mathematically so there is no loss of quality. Every resize is a new original image. Photoshop is a raster driven program based on pixels. The original size is the only size with maximum quality. Any resizing will result in some quality loss.
--
Tom

The camera doesn't make a bit of difference. All of them can record what you are seeing. But you have to see.

http://www.flickr.com/photos/25301400@N00/

In fact, almost all adjustments in Photoshop cause a loss in image data, not just resizing. This is one reason for using adjustment layers and editing 16bit images - not that either of these would help with resizing.

--
Greg

When you've got a moment, have a look at my newly updated site including my blog:
http://www.wrightphotos.co.uk
also http://www.wrightphotos.co.uk/FromeInFocus

Winner of the South West Rural section of the BBC's Picture of Britain Competition.

tbcass said:

The difference you see is due to the difference between Rastor and
Vector. Vector drawings are mathematical representations of all the
shapes in the drawing. When you resize a vector drawing the image is
redrawn mathematically so there is no loss of quality. Every resize
is a new original image. Photoshop is a raster driven program based
on pixels. The original size is the only size with maximum quality.
Any resizing will result in some quality loss.

Click to expand...

Exactly. But Cha Cha seems to be claiming that one can convert a digital camera image (bitmap) into a vector drawing, resize without loss of detail, convert back to bitmap then sharpen, etc., at the reduced size. I maintain that the first step (bitmap --> vector) is impossible.

-- Bill
http://billw.smugmug.com

You can convert bitmap to vector...the process losses a lot of the batmap info.
Then when you convert back to bitmap you lose more.

You only do this to have the outline of the photo, then you can alter the size as big as you want.
That's my understanding.

Ralph.

Celtic Yob said:

You can convert bitmap to vector...the process losses a lot of the
batmap info.
Then when you convert back to bitmap you lose more.
You only do this to have the outline of the photo, then you can alter
the size as big as you want.
That's my understanding.

Ralph.

Click to expand...

I agree. I should have qualified my statement as saying it was impossible to convert a bitmap to vector while retaining all the detail and nuances of the original* .

This is not a method that can be used to losslessly downsize images from a digital camera.

-- Bill
http://billw.smugmug.com

save option. I just crop/downsample to my needs - 960x720 pixel for web purposes and for printing, to whatever minimum resolution my printer vendor needs. Been printing up to 20x30 and lots of 11x14, 12x18 at Costco with displayable results : ).
cheers,
gil
--
**************
I've minimized dreaming the Mark series with the A700
I don't give opinions, just what I see : ).
Now 99.9% jpg but still 100% hand held,
No baits, calls and tricks but will use luck.

All your comments are welcomed because I am really not a photoshop expert; so if you know anything more please continue to challenge. But this whole process began for me due to an article I read in a recent photo magazine (I will have to look up name and edition). But the writer was comparing different resample programs for enlargements and reductions. The program that was most intriguing and effective in his opinion was one that first converted the bitmap photo image into a vector image. He was amazed that this program was able to enlarge to mural size with the least amount of detail loss of all the programs.

So when I read the vector chapter in my "Real World Photoshop" book (which I mentioned in my original post) that compared vector resizing to raster (bitmap) resizing, I immediately recalled the resizing software article that I had read and and decided to experiment.

My procedure was as follows: I imported an a100 jpeg image from the alpha picture motion program into CorelDraw. Naturally the image imported at 52x36" (or whatever the standard size is from the a100). I then downsampled the image by simply resizing it to 15x10" or thereabouts. I then saved the resized jpeg as a tiff (lossless) file into My Documents (windows xp). Then I opened the same jpeg image from Picture Motion (not the tiff version from My Documents) into Photoshop 7. There I downsampled the image, using bicubic, to 15x10".

Then I opened the CorelDraw Tiff from My Documents into Photoshop so that I could compare them side by side. To my astonishment the CorelDraw Tiff had all the original detail whereas the PHotoshop resize lossed detail. As you know this is because a bitmap image must "throw away" pixel data in order to downsize, whereas a vector image is mathematically based. The Photoshop difference amounted to such important details as loss of corduroy (sp?) texture on a sweater, zipper detail was softened, hair strand sharpness was lossed, facial wrinkles were softened and some were lossed, etc.

Following my "zoomed" observations, I made comparison prints using my Epson 7600 printer on Premium Luster Paper. Both prints were given the same unsharp masking at 64/3/2. The detail difference was visible in the 10x15" prints of Beach Front scenes consisting of rocks, homes, water, trees and sky and of portraits and still lifes. I performed this experiment on three or four A100 images.

I felt that this was an important topic for our Sony forum because we are almost always downsizing from our large camera files and because the data loss is most conspicuous when downsizing. But I would really like to hear the insights of you more experienced photographers.

Thank you,
ChaCHa

Making a photo into a vector image will not work very well unless you want something that looks similar to a watercolor painting. The following is what I understand about bitmap vs vector images. If I am missing something here I am happy to have some one correct me.

Bitmap/raster images are based on a grid of color pixels. Each pixel is assigned a specific location and color value. In working with bitmap images, you edit groups of pixels rather than objects or shapes. Because bitmap graphics can represent subtle gradations of shade and color, they are appropriate for photographs.

Vector graphics are made up of lines and curves defined by mathematical objects. They retain their sharpness when they are resized. However, they are not proficient at displaying subtle color gradations. Thus, when you convert a bitmap image with a high level of detail (a photograph) to a vector image (graphic) the result is something like a watercolor painting... at least as far as I know.

By the way, I sometimes convert my JPEG images to vector graphics in Adobe Illustrator via the "live trace" option under the "Object" menu. By utilizing the various tools in Illustrator it is possible to create some very interesting graphics derived from photographs. Illustrator is the Adobe program that provides the most versatility when working with vector images.

Cha Cha,

I've worked with vector programs for over 15 years. I started with Corel Draw, when most of the graphic programs were Mac only, and continued through v5. I moved over to Macromedia Freehand, and still use v8 for my vector work.

As others have pointed out, raster files need to be "traced" to have a facsimile vector image. The many vector programs allow you to import a raster file, and use the raster along with vector shapes.

What you did, was to import a raster file, in this case a tiff, and export the same file to photoshop. Unless traced, the file was never a vector image, although Corel Draw is a vector program.

I'm pasting a link to a truly great vector artist, Bert Monroy. The images are all vector, with additional shading and texture work in Photoshop.

http://www.bertmonroy.com/fineart/text/fineart_damen.htm
--
Regards,
Graham

Best wide-angle lens? Two steps backward. Look for the 'ah-ha'.
Ernst Haas

Cha Cha as far as I know if you import a jpeg into Corel Draw (I have Corel Draw 8 and also the latest Corel X3) and save it as a TIFF it's still only a bitmap, not a vector, you will only get a vector if you save it as a .CDR file, and I don't think that would improve the picture at all, (actually I don't think it's even possible).

It might pay to check the quality level setting in photoshop (if you havn't already done so), that can affect the image detail when you resize and save...... just a thought.
Greg

chacha1 said:

But this whole process began for me due to an article I read in a
recent photo magazine (I will have to look up name and edition).

Click to expand...

Please do. It will help us understand what you are trying to do.

chacha1 said:

But
the writer was comparing different resample programs for enlargements
and reductions. The program that was most intriguing and effective
in his opinion was one that first converted the bitmap photo image
into a vector image. He was amazed that this program was able to
enlarge to mural size with the least amount of detail loss of all the
programs.

Click to expand...

It will help tremendously to know that name of the program that converts a bitmap into a vector image without loss of detail.

chacha1 said:

So when I read the vector chapter in my "Real World Photoshop" book
(which I mentioned in my original post) that compared vector resizing
to raster (bitmap) resizing, I immediately recalled the resizing
software article that I had read and and decided to experiment.

My procedure was as follows: I imported an a100 jpeg image from the
alpha picture motion program into CorelDraw.

Click to expand...

As Greg has explained, at this point you DO NOT have a vector image - you have a bitmap. CorelDraw supports both vector-based drawing and bitmaps. Simply importing a jpeg into CorelDraw DOES NOT convert it into a vector image.

chacha1 said:

Naturally the image
imported at 52x36" (or whatever the standard size is from the a100).
I then downsampled the image by simply resizing it to 15x10" or
thereabouts.

Click to expand...

It would help here to know exactly what options were specified in the CorelDraw resize dialog - especially the equivalent of the photoshop "resample image" checkbox.

chacha1 said:

I then saved the resized jpeg as a tiff (lossless) file
into My Documents (windows xp). Then I opened the same jpeg image
from Picture Motion (not the tiff version from My Documents) into
Photoshop 7. There I downsampled the image, using bicubic, to 15x10".

Click to expand...

And in PhotoShop, obviously the "resample image" checkbox was checked.

chacha1 said:

Then I opened the CorelDraw Tiff from My Documents into Photoshop so
that I could compare them side by side. To my astonishment the
CorelDraw Tiff had all the original detail whereas the PHotoshop
resize lossed detail. As you know this is because a bitmap image
must "throw away" pixel data in order to downsize, whereas a vector
image is mathematically based. The Photoshop difference amounted to
such important details as loss of corduroy (sp?) texture on a
sweater, zipper detail was softened, hair strand sharpness was
lossed, facial wrinkles were softened and some were lossed, etc.

Click to expand...

Something is amiss here. Your CorelDraw version of the original jpeg was a bitmap, not a vector drawing, and when you resized it in CorelDraw you threw away pixels, just like you do when you resize in PhotoShop.

chacha1 said:

Following my "zoomed" observations, I made comparison prints using my
Epson 7600 printer on Premium Luster Paper. Both prints were given
the same unsharp masking at 64/3/2. The detail difference was
visible in the 10x15" prints of Beach Front scenes consisting of
rocks, homes, water, trees and sky and of portraits and still lifes.
I performed this experiment on three or four A100 images.

Click to expand...

Sounds like you wasted a lot of paper and ink.

chacha1 said:

I felt that this was an important topic for our Sony forum because we
are almost always downsizing from our large camera files and because
the data loss is most conspicuous when downsizing. But I would
really like to hear the insights of you more experienced
photographers.

Click to expand...

I'm going to go out on a limb and take a wild guess here. That 52x36 inch size that CorelDraw reports is based on an output resolution of 72 dpi. When you "resized" the image in CorelDraw it did not actually change the image at all - it simply adjusted the resolution so the actual pixels in the image fit in the new size. So it changed the resolution to something around 250 dpi.

In PhotoShop, however, when you enter new image dimensions in the "Image Size" dialog box, AND the "Resample Image" checkbox is checked, Photoshop actually reduces the number of pixels in the image to fit the new output size at the given resolution.

So what you have created is: from CorelDraw a 15x10 inch image at 250 dpi, and from PhotoShop a 15x10 inch image at 72 dpi. So of course when you print them, the CorelDraw version will look better. It should - it's got over three times the resolution!

This has absolutely nothing to do with CorelDraw being a "vector" graphics program, or resizing vector images.

It would really help if you could post the two images: the "resized" CorelDraw image and the "resized" Photoshop image. Or at the very least post the EXIF data from the two images - specifically the horizontal and vertical pixel dimensions and the dpi.

-- Bill
http://billw.smugmug.com

Show us 100% crop

and why you need downsize for printing?
--
pbase supporter http://www.pbase.com/qwz/

The article that I referenced which compared image resizing programs was found in a "Shutterbug Digital Photography How To Guide" special edition magazine for 2007-08. The article was entitled "Increasing Image Resolution For Making Bigger and Better Prints" by Darryl Nicholas (I hope it's okay that I reference his article). The program that changes the bitmap image into a vector is called Imagener Unlimited. There's three versions and the most advanced version is the one that he felt was better than all other programs that he reviewed.

Those posters who claim that CorelDraw is not changing the raster image into a vector image will be interested in learning Daryl's reply to my email inquiry. I wrote Darryl and asked that since CorelDraw uses vector images shouldn't it be converting the bitmap files into vectors before opening onto the graphic page? He replied by saying that, though he does not use vector programs, he was not aware that the line art programs like Illustrator and CorelDraw actually did convert rasters to vectors. So his reply would support the posters to this thread who have been saying that CorelDraw is actually not opening photographs as Vectors, but is actually opening them as bitmaps.

So the overwhelming opinion seems to be that CorelDraw is increasing the saved dpi of the resized image to 250 dpi, while Photoshop is saving the resized file at 72 dpi, which would explain the difference in detail sharpness between the two programs. Actually when I export the CorelDraw file as a tiff (you cannot save as a tiff in CorelDraw8, you must export), I'm saving as a 300 dpi image.

Thank you for your replies and for steering me into a better direction. I'm definitely going to have to read the relevant chapters in my PHotoshop book with more depth. Incidently at no time does the "REal World Photoshop" authors prescribe the work around that I described in my original post, nor do they claim that Draw programs open photographs as vector files. Resizing in CorelDraw was my own theory that I surmised from "REal World's..." vector discsussion. Their discussion of Bitmap resampling is very deep and complicated and I definitely need to give it a more thorough reading.

Incidently, is it possible to save a resized photoshop image at more than 72 dpi. Is there a checkbox where I can select the resolution of the final save?

Thanks again for your help. I'm really thankful that I posted the original thread. Otherwise I would not have had this cleared up.

ChaCHa

PS: Sorry that I cannot post the images requested because I do not have a web site to upload them to before posting.

I just resized one of the photos in Photoshop, but this time I adjusted the dpi to 300, the same dpi that I used to export the CorelDraw resized tiff. This time when I compared the two images, there was no difference in detail. In fact, unless it was my imagination, some areas of the photoshop image had slightly sharper detail. So apparently in my first tests I did not know that photoshop was saving the resized files at 72 dpi, versus the CorelDraw save of 300 dpi. Hence the softer detail. Thank you all for your help. Maybe the web host should delete this thread so that newbies will not read it. I would hate for someone new to Photoshop to not read the thread to the happy ending.

ChaCha

Doesn't matter ChaCha, it's all good fun, (until someone gets hurt).
Greg

Sorry mate,
It's just a lots of mumbo jumbo you're talking about, and all is incorrect.

I suggest you enrolling on some courses about digital photography postprocessing to learn more about images, Photoshop, etc.

===
Tom

Its the best tecnique to transform a bitmap picture in a mathematical representation and then resize ad infinitum...

a "kind of" vectors, fractals technique and software exists for years

Regards,
Luis

ChaCha ,

When i resize any photo for printing on a certian paper size , I want to prevent resampleing and even prevent changing the pixel size .

So I open the Image size box , then i make sure i know the exact pixel dimensions first so i always make sure this is the same before clicking OK .

Then I change the resolution number either by calculating the amount of change required , or by trial and error typing directly in the new resolution DPI which results in the paper size I want to print on , and the exact pixel size I started with .

If the pixel size is not the same I do not click the OK box .

If I forget the original Pixel size I close the box with the X and reopen it new .

You can enter any decimal number needed to get the same pixel size .

In my opinion this results in no loss of image data or quality .