Daniel Reetz, the founder of the DIY Book Scanner community, has recently started making videos of prototyping and shop tips. If you are tinkering with a book scanner (or any other project) in your home shop, these tips will come in handy. https://www.youtube.com/channel/UCn0gq8 ... g_8K1nfInQ

Cameras for the Hackerspace Scanner

Everything camera related. Includes triggers, batteries, power supplies, flatbeds and sheet-feeding scanners, too.
User avatar
Shyamasundara
Posts: 15
Joined: 29 May 2012, 09:43
E-book readers owned: I use PDF on my Mac
Number of books owned: 3500
Country: India
Location: Bangalore, India

Re: Cameras for the Hackerspace Scanner

Post by Shyamasundara » 25 Dec 2013, 10:32

spamsickle wrote:
Shyamasundara wrote:I already have 2 Canons so I will try CHDK, but for me it was not so much the price point but the higher resolution 20 MP compared to 16 MP which give you much better images to start with. Will still go to the Nikon showroom to see what they have to offer.
Unless you are scanning newspapers or something similarly large I would question the value of a 20 MP camera for this application. A standard-sized book can be more than adequately captured with an 8 MP or 10 MP camera. Beyond that, you're increasing your processing time and storage requirements without significantly increasing the quality of the final product. Even small type (superscripts in footnotes, for instance) can be easily resolved in a 12 MP image.
Currently I am using a specialized Plustek bookscanner and scanning in at 600 DPI for BW and 300 DPI for Color. Been doing this for 4 year and wore out a scanner already and on second one, will probably need a 3rd one soon. (I hire someone to scan in my personal library 6 days/week.)

I can definitely notice a difference between an image that is scanned at 300 DPI and 600 DPI even after applying Adobe "clear scan." A page scanned in at 600 DPI and then "clear scanned" is just better and easier on the eyes. And since I am the one who is going to be reading the texts (and I have several thousand PDFs documents) I want the best images so that I don't stress my aging eyes anymore than necessary. ;)

I currently have a 12 MP and 14 MP Canon and will experiment with them and see what kind of out put they generate. But I notice when looking at an image of a sheet of paper taken with the 14 mp that it only had a resolution of 180 pixels/inch. To me that seems pretty low.

But I am willing to experiment and see what kind of final product I get.

As for the price, the Nikon 20 MP are about $90-$110/each, very affordable for a camera that would just be used for scanning.

Anyway, this is all hypothetical and can only be decided after I actually do some experimentation. My main concern is to get the highest resolution image possible for feeding to Adobe "clear scan." It makes a difference to my eyes as I read these documents for hours at a time daily. Not just a little light reading now and then.

spamsickle
Posts: 596
Joined: 06 Jun 2009, 23:57

Re: Cameras for the Hackerspace Scanner

Post by spamsickle » 25 Dec 2013, 14:47

I hire someone to scan in my personal library 6 days/week.
I'm too cheap to do that, but I envy you anyway.
I notice when looking at an image of a sheet of paper taken with the 14 mp that it only had a resolution of 180 pixels/inch. To me that seems pretty low.
Seems low to me too. I have been using old 8 MP cameras to scan my books for years, and precede each scan with a few shots of the book and a tape measure. I just took a look at the most recent scan, and I have 9.375 inches in the frame in the horizontal direction, which is 3264 pixels. That works out to a bit more than 340 pixels per inch by my calculations. If you're only getting 180 pixels/inch with a 14 MP camera, you may not be filling as much of the frame with document as you could be.
A page scanned in at 600 DPI and then "clear scanned" is just better and easier on the eyes. And since I am the one who is going to be reading the texts (and I have several thousand PDFs documents) I want the best images so that I don't stress my aging eyes anymore than necessary. ;)
...
My main concern is to get the highest resolution image possible for feeding to Adobe "clear scan." It makes a difference to my eyes as I read these documents for hours at a time daily. Not just a little light reading now and then.
I guess your eyes and mine work differently. I spend most of my day reading a computer screen with text from websites like this one. I just got out the magnifying glass, and the characters I'm reading right now are 8x8 pixels for an "a" or an "e", presumably a few more for "k"s and "g"s and capitals. Compared to that, the (roughly) 32x32-pixel equivalent I get in a bookscanned image is practically decadent. I've never noticed any eyestrain; I guess I'm just lucky.

I don't think it's necessarily an age thing; my offspring doesn't like to read anything but paper or Kindle for any length of time. If you notice a difference, and it matters to you, then definitely go for what meets your needs.

User avatar
Shyamasundara
Posts: 15
Joined: 29 May 2012, 09:43
E-book readers owned: I use PDF on my Mac
Number of books owned: 3500
Country: India
Location: Bangalore, India

Re: Cameras for the Hackerspace Scanner

Post by Shyamasundara » 26 Dec 2013, 02:03

spamsickle wrote:
I hire someone to scan in my personal library 6 days/week.
I'm too cheap to do that, but I envy you anyway.
It only costs me about $100/month. 8-) But those scanners are way too slow and I have a lot more books to scan.

spamsickle wrote:
I notice when looking at an image of a sheet of paper taken with the 14 mp that it only had a resolution of 180 pixels/inch. To me that seems pretty low.

Seems low to me too. I have been using old 8 MP cameras to scan my books for years, and precede each scan with a few shots of the book and a tape measure. I just took a look at the most recent scan, and I have 9.375 inches in the frame in the horizontal direction, which is 3264 pixels. That works out to a bit more than 340 pixels per inch by my calculations. If you're only getting 180 pixels/inch with a 14 MP camera, you may not be filling as much of the frame with document as you could be.
Maybe I got it wrong, I just opened up the image in PhotoShop and it said that the image had a resolution of 180 DPI. The image did not tightly hug the page and there was a slight border. But that should not make that big of a difference. So it seems that PS's DPI and pixels is not the same.


spamsickle wrote:
A page scanned in at 600 DPI and then "clear scanned" is just better and easier on the eyes. And since I am the one who is going to be reading the texts (and I have several thousand PDFs documents) I want the best images so that I don't stress my aging eyes anymore than necessary. ;)
...
My main concern is to get the highest resolution image possible for feeding to Adobe "clear scan." It makes a difference to my eyes as I read these documents for hours at a time daily. Not just a little light reading now and then.
I guess your eyes and mine work differently. I spend most of my day reading a computer screen with text from websites like this one. I just got out the magnifying glass, and the characters I'm reading right now are 8x8 pixels for an "a" or an "e", presumably a few more for "k"s and "g"s and capitals. Compared to that, the (roughly) 32x32-pixel equivalent I get in a bookscanned image is practically decadent. I've never noticed any eyestrain; I guess I'm just lucky.

I don't think it's necessarily an age thing; my offspring doesn't like to read anything but paper or Kindle for any length of time. If you notice a difference, and it matters to you, then definitely go for what meets your needs.
When you read text on websites like this you are not looking at bitmap images that pixilate if you enlarge the fontsize.

Adobe "clear scan" does create vector based fonts from the bitmap images but naturally if you supply better quality input you get higher quality output. That has been my experience.

cday
Posts: 295
Joined: 19 Mar 2013, 14:55
Number of books owned: 0
Country: UK

Re: Cameras for the Hackerspace Scanner

Post by cday » 26 Dec 2013, 07:58

Scanning at a higher resolution gives ClearScan a better chance of accurately reproducing the original font which, given the subtlety of font design, leads to a more pleasing appearance:
ClearScan.png
The above image is taken from this Adobe link: http://blogs.adobe.com/acrolaw/2009/05/ ... n_is_smal/

Scanning at a higher resolution is also likely to reduce the problem which has been reported in the past of ClearScan sometimes producing a very large number of fonts, likely due to not recognising different instances of lower resolution characters as being part of the same font.

User avatar
Shyamasundara
Posts: 15
Joined: 29 May 2012, 09:43
E-book readers owned: I use PDF on my Mac
Number of books owned: 3500
Country: India
Location: Bangalore, India

Re: Cameras for the Hackerspace Scanner

Post by Shyamasundara » 26 Dec 2013, 12:39

cday wrote:Scanning at a higher resolution gives ClearScan a better chance of accurately reproducing the original font which, given the subtlety of font design, leads to a more pleasing appearance:

Scanning at a higher resolution is also likely to reduce the problem which has been reported in the past of ClearScan sometimes producing a very large number of fonts, likely due to not recognising different instances of lower resolution characters as being part of the same font.
Thank you for posting that image, a picture is definitely worth a 1000 words. With high quality input I have found that "ClearScan" can produce beautiful looking output with the added advantage of reducing the PDF by 80% or more. And no matter how much you magnify the pdf it doesn't pixilate.

The only time there is no size reduction is if you have a text with different languages like English, Sanskrit, Arabic, Persian, Chinese and Greek. I have texts that often contain English and Sanskrit and others with 4 or more different languages that do not use a Latin font. In such cases the file will become larger, sometimes significantly larger. But the increased readability is worth it.

The other time there is size no reduction is, as Cday noted, if you start with poor quality images in your pdf.

That is the reason I want to use cameras that give the maximum amount of resolution for high quality output and readability.

User avatar
Shyamasundara
Posts: 15
Joined: 29 May 2012, 09:43
E-book readers owned: I use PDF on my Mac
Number of books owned: 3500
Country: India
Location: Bangalore, India

Re: Cameras for the Hackerspace Scanner

Post by Shyamasundara » 07 Jan 2014, 03:30

spamsickle wrote:
I notice when looking at an image of a sheet of paper taken with the 14 mp that it only had a resolution of 180 pixels/inch. To me that seems pretty low.
Seems low to me too. I have been using old 8 MP cameras to scan my books for years, and precede each scan with a few shots of the book and a tape measure. I just took a look at the most recent scan, and I have 9.375 inches in the frame in the horizontal direction, which is 3264 pixels. That works out to a bit more than 340 pixels per inch by my calculations. If you're only getting 180 pixels/inch with a 14 MP camera, you may not be filling as much of the frame with document as you could be.
I found out what the problem was, why I was only getting 180 DPI. Even though the original is a piece of paper approx 11.5 X 8, when I checked the size of the image in PhotoShop it was about 22.2 X 16.5 about double the size. I could change the setting in photo shop to double the DPI and halve the dimensions by creating an automation/action. But that just adds another step.

Is there a way of making the image the correct size in the first place?

spamsickle
Posts: 596
Joined: 06 Jun 2009, 23:57

Re: Cameras for the Hackerspace Scanner

Post by spamsickle » 07 Jan 2014, 06:36

I don't think there's any way to do it without adding a step. The problem is that a camera really has no good way to tell what a DPI value should be. If you're taking a macro picture of two ants fighting, you may have thousands of pixels per inch. If you're taking a picture of a distant galaxy, you may have trillions of miles per pixel. I suppose in theory it might be possible for the camera software to estimate ballpark values by accounting for focus distance (if less than infinity) and zoom level, but I think most (all?) don't bother and simply use a constant value for the EXIF DPI value.

A scanner knows what an inch is in the images it produces - it either moves the scanning head an inch, taking a certain number of samples as it does so, or feeds a sheet an inch. The resolution of its scanning head is fixed. The EXIF DPI value is only meaningful if you're using a real scanner, or you're going in after the image is shot and setting it based on your knowledge of what's in the picture.

In most cases, the EXIF DPI doesn't matter. If you're planning to print the images, you may want to make it accurate if your printer software is using it. Scan Tailor asks for, and allows you to set, the number.

mhhelle
Posts: 7
Joined: 01 Jan 2014, 12:25
E-book readers owned: kindle
Number of books owned: 200
Country: USA

Re: Cameras for the Hackerspace Scanner

Post by mhhelle » 10 Jan 2014, 09:27

I’m trying to decide on cameras for the hackerspace build. I read through the camera guide spreadsheet and other threads.

I’m looking at refurbished Canon Powershots and compared the A810, A2300, A2400IS, A3400 and A4000IS. They no longer have the A2200 for sale from Canon. It is on amazon but $100 so seems not worth it to me.

- They all have 16 megapixels
- A2400IS, A3400, A4000IS all have image stabilization. A810 and A2300 do not
- The A4000IS has a better zoom (8x optical versus 5x for all the others)
- chdk is alpha on A2300, A2400IS, A3400, A4000IS but seems to be fully supported on A810
- Differences in LCD screen (doesn’t matter)
- Differences in video capabilities (doesn’t matter)
- A810 takes AA batteries and uses a different power adapter
- Differences in size/weight/dimensions (doesn’t matter)
- Price (refurbished): A810: $59.99, A2300: $74.99, A2400IS: $79.99, A3400: $89.99, A4000IS: $109.99

I’m guessing the zoom won’t matter much? If that’s true then I think it’s down to the A810 versus the A2400IS for me. Question is—is the image stabilization worth $20 per camera over the A810, and will I encounter more difficulties if chdk is alpha on that camera?

User avatar
daniel_reetz
Posts: 2797
Joined: 03 Jun 2009, 13:56
E-book readers owned: Used to have a PRS-500
Number of books owned: 600
Country: United States
Contact:

Re: Cameras for the Hackerspace Scanner

Post by daniel_reetz » 10 Jan 2014, 17:51

Right now I have the A2200 and am working on the A1400 (very cheap at the moment). Those or the A810 seem reasonable to me. A2200 has the widest adoption of any camera. A1400 may follow it.

malcolmk
Posts: 1
Joined: 10 Jan 2014, 12:33
E-book readers owned: none
Number of books owned: 400
Country: England

Re: Cameras for the Hackerspace Scanner

Post by malcolmk » 11 Jan 2014, 14:21

I've just bought a second hand A810 for 20 UKP (UK pounds). SDM installed without any problems and appears to work - it took a photo when I plugged the usb into my laptop, then unplugged it. I'm trying to understand the menus and scripts in SDM before powering it from a switch mode power supply - batteries don't seem to last very long.
Malcolm

Post Reply