09-30-2011, 08:50 AM
(This post was last modified: 09-30-2011, 08:57 AM by Brightcolours.)
[quote name='boren' timestamp='1317340770' post='11983']
So fill in the blanks then - what DPI needs to be defined for each print to make the two comparable? And what printer supports such fine settings?
That's why I wrote "resample" and not "downsize".
What printer do you suggest for this test? What DPI settings?
Because A4 is the largest size that most home printers support. If you have a specific alternative methodology in mind, feel free to share it! My hunch is that whatever you're going to suggest will be far too costly to be of use to people who want to compare the output of camera A with that of camera B. This is why normalizing the size of files and comparing on screen side by side is what most people prefer to do, at least those who understand why comparing 100% crops is useless.
I don't know what you mean by "normalize before measuring". I'm referring to normalizing the size of the resulting files. For example, upsizing K5 images to match the A77 resolution and comparing them with the native Sony images, and to make the test complete do it the other way around - compare downsized A77 images with native K5 ones.
[/quote]
WE were originally talking about downsizing to normalize. That you started to talk about upsizing is your doing.
We were also talking about a rest site/company/organization like DXOmark. Not you at home with some simple A4 printer. So again, not related to what we were discussing.
On DPI's crazy to think about DPIs. Like I said, 300 and 600 DPI is from the black and white post script printer era. 1980's to mid 1990's.
Dots are not pixels, pixels are not dots.
Printers have DPI's of upto 4000 or so. Those dots have a single primary ink colour, more dots are needed to create a coloured pixel of an image. But a pixel can start anywhere on that 4000 dots scale. So yes... no problems there in the printer influencing noise when just printing full page without worrying about any DPI setting. Just print the image as is at pull page, no resampling to be done. Just let the printer decide on how many dots it uses for each pixel. Not hard, not complex, no upscaling or downscaling image pixel count itself.
So dont touch the files themselves. Get it out of your head that you have to up or downscale the files, before printing. Just print the images the same size, nothing to it.
WE were talking about DXOmark normalizing before measuring. So, downsampling images to a varying degree to end up with similar sized images for different sensors, and that being problematic according to me. So, normalizing before measuring.
As it turns out, that is NOT what DXOmark does. They normalize the results AFTER measuring.
And that is not about printing. If it is about printing the same size, just dont upsample or downsample the images... just print them the same size! Nothing complex about that, nothing problematic about that. Printers have high enough resolution for that, and smart enough software to calculate pixel size correctly for each file from each camera.
So fill in the blanks then - what DPI needs to be defined for each print to make the two comparable? And what printer supports such fine settings?
That's why I wrote "resample" and not "downsize".
What printer do you suggest for this test? What DPI settings?
Because A4 is the largest size that most home printers support. If you have a specific alternative methodology in mind, feel free to share it! My hunch is that whatever you're going to suggest will be far too costly to be of use to people who want to compare the output of camera A with that of camera B. This is why normalizing the size of files and comparing on screen side by side is what most people prefer to do, at least those who understand why comparing 100% crops is useless.
I don't know what you mean by "normalize before measuring". I'm referring to normalizing the size of the resulting files. For example, upsizing K5 images to match the A77 resolution and comparing them with the native Sony images, and to make the test complete do it the other way around - compare downsized A77 images with native K5 ones.
[/quote]
WE were originally talking about downsizing to normalize. That you started to talk about upsizing is your doing.
We were also talking about a rest site/company/organization like DXOmark. Not you at home with some simple A4 printer. So again, not related to what we were discussing.
On DPI's crazy to think about DPIs. Like I said, 300 and 600 DPI is from the black and white post script printer era. 1980's to mid 1990's.
Dots are not pixels, pixels are not dots.
Printers have DPI's of upto 4000 or so. Those dots have a single primary ink colour, more dots are needed to create a coloured pixel of an image. But a pixel can start anywhere on that 4000 dots scale. So yes... no problems there in the printer influencing noise when just printing full page without worrying about any DPI setting. Just print the image as is at pull page, no resampling to be done. Just let the printer decide on how many dots it uses for each pixel. Not hard, not complex, no upscaling or downscaling image pixel count itself.
So dont touch the files themselves. Get it out of your head that you have to up or downscale the files, before printing. Just print the images the same size, nothing to it.
WE were talking about DXOmark normalizing before measuring. So, downsampling images to a varying degree to end up with similar sized images for different sensors, and that being problematic according to me. So, normalizing before measuring.
As it turns out, that is NOT what DXOmark does. They normalize the results AFTER measuring.
And that is not about printing. If it is about printing the same size, just dont upsample or downsample the images... just print them the same size! Nothing complex about that, nothing problematic about that. Printers have high enough resolution for that, and smart enough software to calculate pixel size correctly for each file from each camera.