Re: Sola Technical's $70 acrylic scanning platform on Ponoko
Posted: 08 Dec 2011, 17:18
After eagerly following the new model scanner development, I've decided to finish my tests with this scanner before proceeding with that one. So, let's buy a camera and build a blackout tent for it:
It's holiday season 2011, you can reasonably get 14MP and 16MP point-and-shoot compact and ultracompact cameras for around $100 or less. Neither this model nor the new kit require automated triggering, so this opens the question of other camera models besides the latest/most affordable Canon.
Imaging Resource (IR) has detailed reviews with image quality analyses for many low-end cameras, unlike DP Review, which seems to focus on high-end ones. Based on historical reviews on DP Review, Amazon user reviews and other forum users' use of CHDK, Canon, Sony and Panasonic cameras seem to be the best bets for reliable camera purchasing, rather than Nikon, Kodak or Fuji.
It also appears that cameras labeled as Bell & Howell, Vivitar and GE are similar model OEM cameras with incredibly poor lenses/sensors/false MP ratings (according to anecdotes on the internet) and should be avoided.
Panasonic Lumix camera lines seem to have similar sensors and features, and are differentiated by their housing and lenses: S series < FP series < FH series.
On all the IR reviews, I'm looking exclusively at sharpness, geometric distortion, chromatic aberration, resolution, ISO (detail at low ISO), performance and conclusion.
Based on that, these seem to be best bets for 16MP cameras:
With the daintiness I suspect will be required handling the platen, picture-taking speed doesn't matter as much, so I've ordered the Sony because the low ISO performance seems to be a little better than the Canon's, and also to be contrary.
For testing performance in the scanner itself, a standardized chart was discussed here on the forums, but I don't understand why we wouldn't just use industry standard ones, similar to the tests IR does. Any chart we actually print at home is going to have a lot of variance in the visible detail and such because we all have different printers with different inks and different paper. I feel like we need to use ones produced by outside sources for consistency, even if it's just having a set of images in a print-on-demand book through Blurb or Lulu or wherever that have been color-matched with them.
IR's tests include an ISO standard chart for testing resolution. This is super interesting to me. The official PDF is ~100 (your unit of currency); in the US that's through ANSI for $142, but I think you really want to buy pre-printed paper charts for around the same price: http://www.precisionopticalimaging.com/ ... type=12233 or http://www.edmundoptics.com/products/di ... uctid=2287
That said, a replica is available here as a PDF, and I'll be printing this out and using it for basic tests: http://www.graphics.cornell.edu/~westin ... chart.html
If more people get to using it, maybe Scan Tailor and Book Scan Wizard could integrate code for doing slant-edge analysis to determine the resolution and help compare performance. Matlab code for that is here: http://www.i3a.org/resources/#iso
Color tests like this might also be nice: http://www.edmundoptics.com/products/di ... uctid=1815
Anyway, the Sony will be here tomorrow, and then I'll take pictures of the blackout tent with it.
It's holiday season 2011, you can reasonably get 14MP and 16MP point-and-shoot compact and ultracompact cameras for around $100 or less. Neither this model nor the new kit require automated triggering, so this opens the question of other camera models besides the latest/most affordable Canon.
Imaging Resource (IR) has detailed reviews with image quality analyses for many low-end cameras, unlike DP Review, which seems to focus on high-end ones. Based on historical reviews on DP Review, Amazon user reviews and other forum users' use of CHDK, Canon, Sony and Panasonic cameras seem to be the best bets for reliable camera purchasing, rather than Nikon, Kodak or Fuji.
It also appears that cameras labeled as Bell & Howell, Vivitar and GE are similar model OEM cameras with incredibly poor lenses/sensors/false MP ratings (according to anecdotes on the internet) and should be avoided.
Panasonic Lumix camera lines seem to have similar sensors and features, and are differentiated by their housing and lenses: S series < FP series < FH series.
On all the IR reviews, I'm looking exclusively at sharpness, geometric distortion, chromatic aberration, resolution, ISO (detail at low ISO), performance and conclusion.
Based on that, these seem to be best bets for 16MP cameras:
- Panasonic Lumix DMC-FH5 - no IR review, but a better-lensed model does - Amazon (out of stock from Amazon, available from other sellers)
- Sony Cyber-Shot DSC-W570 - IR review - Amazon
- Canon Powershot A3300 - IR review - Amazon
With the daintiness I suspect will be required handling the platen, picture-taking speed doesn't matter as much, so I've ordered the Sony because the low ISO performance seems to be a little better than the Canon's, and also to be contrary.
For testing performance in the scanner itself, a standardized chart was discussed here on the forums, but I don't understand why we wouldn't just use industry standard ones, similar to the tests IR does. Any chart we actually print at home is going to have a lot of variance in the visible detail and such because we all have different printers with different inks and different paper. I feel like we need to use ones produced by outside sources for consistency, even if it's just having a set of images in a print-on-demand book through Blurb or Lulu or wherever that have been color-matched with them.
IR's tests include an ISO standard chart for testing resolution. This is super interesting to me. The official PDF is ~100 (your unit of currency); in the US that's through ANSI for $142, but I think you really want to buy pre-printed paper charts for around the same price: http://www.precisionopticalimaging.com/ ... type=12233 or http://www.edmundoptics.com/products/di ... uctid=2287
That said, a replica is available here as a PDF, and I'll be printing this out and using it for basic tests: http://www.graphics.cornell.edu/~westin ... chart.html
If more people get to using it, maybe Scan Tailor and Book Scan Wizard could integrate code for doing slant-edge analysis to determine the resolution and help compare performance. Matlab code for that is here: http://www.i3a.org/resources/#iso
Color tests like this might also be nice: http://www.edmundoptics.com/products/di ... uctid=1815
Anyway, the Sony will be here tomorrow, and then I'll take pictures of the blackout tent with it.