Resolution - and quality
Posted 22/01/2004 - 00:13
Link
Sorry folks,
The previous questions, image size and Linear Processing, should have had my name attached.
The previous questions, image size and Linear Processing, should have had my name attached.
Lurach
Posted 22/01/2004 - 04:59
Link
Hi Lurach,
What a big question! Sorry to confuse you with my threads
First, the filesize issue...
If you have a large image and print it small, the printer driver has to downsample to produce the output. Obviously it depends on the output technology and DPI. If you are outputting to film, say at 310dpi, then it's pretty obvious what has to happen. In fact, I find that doing an image resize and applying some sharpening within PS is best - that way I can print 1:1 to the driver (in my case the HiTi Photoshuttle).
When printing to inkjet things get a little more complicated. The inkjet DPI value is not easily related to optical/photographic DPI since the drops of ink are effectively binary and have to be 'dithered' to produce the appearance of continuous tone. Personally, I find that an image of around 300dpi (resize to 8"x10" at 300dpi = 2400pixels x 3000pixels) works really well with the Epson inkjet.
So, yes, some information has to be thrown away - it is physically impossible not to (unless you had an infinite DPI output ) Just where that information gets thrown away is somewhat down to your workflow.
Now to the subject of scanners...
When I had the Scan Dual SCSI I found that the Minolta driver was horrible. It produced yucky posterised images and was inconsistent. I switched to Vuescan and have always found that so very reliable for scanning slides and negatives - no posterisation, absolute control, and very very good for batch scanning a roll of film. (Almost no user intervention required - shove slides in and autoscan, then sort out the images later.)
You may find linear output tricky in Corel. It can be done, but requires setting up colour management which I find more confusing than Photoshop. First off, get your monitor calibrated - http://www.praxisoft.com and download Wiziwyg. It takes ten minutes to calibrate your monitor.
Linear colour space is well suited to handling images from CCDs (digital cameras and scanners included). In my threads I was handling 16-bit per channel (48bit image) files in linear colour space. I'm on a steep learning curve at the moment trying to piece it all together. However, I reckon it will be a bit nasty trying to do 8bit per channel linear colour - there's probably not enough information and you will get severe posterisation in the shadows.
Just to recap... before I switched to Photoshop CS I found that the combination of Wiziwyg, Vuescan, and Corel Photopaint produced very good and repeatable results. In addition, when printing I let Photopaint and the Epson drivers sort it out (but do read the paper handling instructions when setting up the driver)
Phew! Hope that helps answer some of those queries.
Matt
What a big question! Sorry to confuse you with my threads
First, the filesize issue...
If you have a large image and print it small, the printer driver has to downsample to produce the output. Obviously it depends on the output technology and DPI. If you are outputting to film, say at 310dpi, then it's pretty obvious what has to happen. In fact, I find that doing an image resize and applying some sharpening within PS is best - that way I can print 1:1 to the driver (in my case the HiTi Photoshuttle).
When printing to inkjet things get a little more complicated. The inkjet DPI value is not easily related to optical/photographic DPI since the drops of ink are effectively binary and have to be 'dithered' to produce the appearance of continuous tone. Personally, I find that an image of around 300dpi (resize to 8"x10" at 300dpi = 2400pixels x 3000pixels) works really well with the Epson inkjet.
So, yes, some information has to be thrown away - it is physically impossible not to (unless you had an infinite DPI output ) Just where that information gets thrown away is somewhat down to your workflow.
Now to the subject of scanners...
When I had the Scan Dual SCSI I found that the Minolta driver was horrible. It produced yucky posterised images and was inconsistent. I switched to Vuescan and have always found that so very reliable for scanning slides and negatives - no posterisation, absolute control, and very very good for batch scanning a roll of film. (Almost no user intervention required - shove slides in and autoscan, then sort out the images later.)
You may find linear output tricky in Corel. It can be done, but requires setting up colour management which I find more confusing than Photoshop. First off, get your monitor calibrated - http://www.praxisoft.com and download Wiziwyg. It takes ten minutes to calibrate your monitor.
Linear colour space is well suited to handling images from CCDs (digital cameras and scanners included). In my threads I was handling 16-bit per channel (48bit image) files in linear colour space. I'm on a steep learning curve at the moment trying to piece it all together. However, I reckon it will be a bit nasty trying to do 8bit per channel linear colour - there's probably not enough information and you will get severe posterisation in the shadows.
Just to recap... before I switched to Photoshop CS I found that the combination of Wiziwyg, Vuescan, and Corel Photopaint produced very good and repeatable results. In addition, when printing I let Photopaint and the Epson drivers sort it out (but do read the paper handling instructions when setting up the driver)
Phew! Hope that helps answer some of those queries.
Matt
http://www.mattmatic.co.uk
(For gallery, tips and links)
(For gallery, tips and links)
Add Comment
To leave a comment - Log in to Pentax User or create a new account.
2277 posts
54 years
Well I thought this was going to be a relatively straight-forward question until I logged in and read the posts on *istD RAW support and Linear Processing. Now I'm not sure if I know where I am .
Well I suppose now I'll have to ask a couple of questions, so first things first, a question regarding resolution and "quality". We all know that displaying an image at a "size" greater than was originally intended does not lead to a good image, i.e. printing a 640x480 at A3 size isn't exactly going to win too many competitions. But what about the opposite situation? Leaving aside questions of file size and disk space, will printing a 3008x2008 Best image at for example 5"x7" size produce any better result than, say, printing a 1536x1024 Best at 5"x7" or is the "surplus" data in such cases just going to be disgarded? In other words if you know what the final output is going to be is "shooting bigger" of any benefit? - apart from more latitude in the future if needed.
Now to a question that was not originall intended but reading about "Linear Processing" prompts it. By sheer coincidence, today I started to use a Minolta Dimage Scan Dual II scanner. I got it a while ago, but only started using it today and, at first, I could not get satisfactory results from any slides I was scanning. I can only describe the on-sreen results as lifeless, flat or"lacking gloss" I then started to do some "controlled" scans selecting or de-selecting various options. (I am using the default Minolta software and am not using any of its' correction facilities, doing all further correcting in Corel Photo-Paint V9.)
The outcome was that I found I had to avoid the 16bit Linear processing option in the Minolta software to obtain a good image. What is Linear Processing and why was it producing "inferior" scans?