Canon A2600 vs Ixus / ELPH 160?

[dnwst]

Hey all,

I've got a DIY book scanner kit heading my way soon from http://www.diybookscanner.eu/ (excited!), but I'm not quite sure which camera to get. I'm choosing between...

- Canon Powershot A2600
- Canon ELPH 160 (or Ixus 160 as it's called here in Europe).

I'm mostly going to scan old text-heavy books with black and white illustrations and sketches (some of which are quite detailed with fine lines).

I see that the kit from Tenrec Builders include the ELPH / Ixus 160, but looking in their user Electronics Guide (http://tenrec.builders/archivist/electronics-guide.html) it seems they used to include the Powershot A2500 once upon a time.

From what I can gather the difference between the Ixus160 and A2600, that might be relevant for book scanning, are:

- resolution, Ixus 160 has 20mp while A2600 has 16mp.
- optical zoom, Ixus 160 has x8 while A2600 has x5
- aperture, Ixus 160 has f/3,2 while A2600 has f/2,8
- chdk support, I can't see that Ixus 160 is a supported model on http://chdk.wikia.com/wiki/CHDK while A2600 is. Is this correct?

If any of you have experience with using any of these two models I'd love to get your thoughts on this. From what I can understand the A2600 is a bit more tried and tested for book scanning purposes. Here's some questions I'm asking myself though:

1. Does any of these two models provide higher quality scans?

2. Is any of them more future proof in terms of software compatibility / CHDK updates? (e.g. the Ixus 160 which seems to be a newer model)

3. Does the extra 4mp bump in resolution matter when scanning only old books?

4. Is there any newer and better sub $150 camera model that's worth consideration?

Thanks for taking your time reading this! (and even more thanks if you reply!)

All the best,
Dan

(I might add that I've read through lots of threads in this forum before posting these questions, but it seems none of these models are much talked about even though they're sold with the standard kit)

[qv_]

- resolution, Ixus 160 has 20mp while A2600 has 16mp.
Depending on the size of the book. I like to have 300dpi at the end.
16MP and size A4 is not realy 300DPI in the end

- optical zoom, Ixus 160 has x8 while A2600 has x5
you need to see "only the page" so if the zoom is "wrong" you can change the mounting of the camera but I think both will work OK

- aperture, Ixus 160 has f/3,2 while A2600 has f/2,8
You will use 5,6 or 8 so is no deal.

- chdk support, I can't see that Ixus 160 is a supported model
No clue

To get 300 DPI you need the best lens and good light.
Many people convert camera sensor pixel direct into DPI but that is not correct. No lens is that good. All lens lose minimus 10% resolutin (and more in the coner).

/Jan

[duerig]

First, a quick clarification. The older camera is the A2500. I have never used an A2600.

(1) The ELPH 160 will provide marginally better scans. It is a newer camera with a slightly more dense sensor. As qv_ mentions, the headline resolution (20MP) is best thought of as a maximum instead of a fixed value. But it should get you around 300 DPI on the platen of the Archivist. You can do a lot better in terms of cameras, but that will cost a lot more money.

(2) They should both continue to work fine indefinitely. The A2500 is getting harder and harder to acquire, which is why my electronics pack now uses the ELPH 160. The advantage of the A2500 is that it is better supported now. Eventually, both will be well supported I hope.

(3) The extra 4mp bump in resolution will get you a little bump in DPI. It is nice, but not super-important. IMO, the only real reason to get an ELPH 160 over an A2500 is that it is easier and cheaper to acquire than the out of production A2500. The difference in optical zoom doesn't matter because you will not be limited by zoom in the Archivist setup. I haven't noticed any difference based on the aperture in test scans.

(4) I don't know of any cheap camera model that is better than the A2500 or ELPH 160 right now.

The ELPH 160 is not yet on the main build. Part of the reason is that one person who was using a test build of CHDK bricked their camera. For details, see this write-up:

http://tenrec.builders/cameras/elph160.html

I have tried very hard to brick a camera to find out for sure if the bug that they found was fixed. Since I have been unable to do this, I can't say for sure that the problem is fixed. But I am fairly sure that using the camera as part of Spreads will not damage the camera. Which is one reason that I have started including the ELPH 160 in the kits. I include my own build of CHDK for the cameras as part of the kit. And I plan on releasing a link to this build along with the warning at some point in the near future.

Basically, the A2500 will be the easiest for you to get working right now and is the best-supported. Which is why I pushed it forward as the 'standard' earlier. But since it is getting harder to acquire, I am hoping that the ELPH 160 will work out as a new 'standard' soon. Right now it works well enough that I am shipping it to people that I can personally support. Soon I hope that the evidence will show that it is a good 'standard' for everyone.

-D

[rolling29]

I recently got the ELPH 160 when it was on sale for $99 and have been testing / playing with it for about a week. Because of that single instance of a reported brick, they have removed public links to the CHDK build. If you join up and ask for it they seem very willing to send links to people that are willing to try it with full acceptance of the risk. In my opinion, I will never run anything close to that scenario so the risk is pretty low. I do not try to have multiple partitions on the SD card. I do not want to use raw format image files. And spreadpi is supposed to save the images without filling up the SD card anyway. The chances of me bricking a device due to running out of space is pretty much zero. I saw duerig's posts over on CHDK and my general respect for his competence also added to the low risk assessment. I tried to brick it in normal usage, but he actually diligently tried to brick it by doing exactly what the bug report described, and couldn't get it to happen. My anecdote doesn't count for much, but I have had no trouble with build "1.4.0-4278-full".

Now, my sort of review of the 160...

When qv_ is talking about resolution it's important to distinguish the meaning. https://en.wikipedia.org/wiki/Image_resolution mentions line pairs per mm, and this is what really matters to me. 10% loss is very generous. A friend of mine works at a university library, and also likes photography. This made for an interesting experiment with my A2200, T2i, the new ELPH 160, and his 6D and shiny new 5DSR. That library also happens to have some Bookeye scanners so we had an "official", if a bit old, test target to play with. Among the many strange areas on this sheet are various LPPM boxes. The line pairs per mm test image results were not what we expected.

To try and keep things fair all cameras were on a tripod shooting the target in the same light. The DSLRs were all using the same 40mm lens, and the P&S cameras used what they had. They pretty much all (except the 5DSR) ended up with the 3.2-3.6 line pair sets clearly resolved, and sometimes up to 4.0 but only at the center. Anything smaller was just a gray blob instead of clear black and white lines. Converting that to DPI you get about 150-200 at best. We set the tripod distance in each case so that image width would come out as close to 300pixels per inch as possible. Turns out it wasn't that hard because the target was a certain size for a reason. Well, except the 5DSR with it's 50MP sensor. We set that at the same width as the other DSLRs and it should have been just over 500 DPI. We got 7.2 line pairs at best, equivalent to about 350 DPI. In all cases reading the text labeled as 6pt was ok, but sometimes difficult. 4pt was too difficult for me to make out the letters. 7pt and up was easy enough to read.

Doing our best to get the clearest images possible, and he had some special filters for long exposures, we still lost 30-50% from pixel to optical resolution. There was one other major thing we noticed besides optical line pair resolution...

Chromatic aberration.
The DSLRs all had nice sharp black and white test patterns all over the place.
The A2200 and EPLH 160 both had really bad green and red shifting along the contrast line out at the target edges. Instead of white-black-white patterns, they produced white-green-black-red-white patterns. We tried everything we could think of, but the color shift just wouldn't go away. (Using CHDK to lock the neutral density filter in place and increase the exposure time did noticeably help.)

Don't let my informal experiment convince you the 160 is a bad camera, that is not my intention. It is important to remember that the 40mm DSLR lens, just by itself, cost 50% more that the entire ELPH 160 camera. In fact, I consider getting a similar 30-50% "lens loss" to be keeping up very well with the better cameras. The chromatic aberration is disappointing, but may not be an issue at all when normally shooting gray-scale images for books, and when using the camera for everyday family pictures there's no precise German test patterns in the background to trick the sensor. In all honesty, I never even noticed it before with the A2200.

I still like my A2200's from ebay for ~$25. When they finally die I'll happily switch to 160's full time. To me the bump in MP just means bigger image files to scale down and process.

[qv_]

Doing our best to get the clearest images possible, and he had some special filters for long exposures, we still lost 30-50% from pixel to optical resolution.

Did you have the light in 45 degree and try different typs of light?
I did notis a 50% difference in LPMM depending on the light type and angel.

/Jan

yes -10% is only with the wery costly primes...

[rolling29]

Did you have the light in 45 degree and try different typs of light?

No, I wish we had some light sources available at the time and place, but the best we could do was keep it uniform between shots. It was daylight coming in from a north-facing window and the target was about perpendicular to it. I guess you could say the light was nearly 45 degrees, but there was a slight but clear gradient away from the window.

Let's see if I can post these test images without making a mess of it...

center 5DSR EF40mm.JPG (113.93 KiB) Viewed 10894 times

center T2i EF40mm.JPG (67.97 KiB) Viewed 10894 times

center ELPH 160.JPG (68.53 KiB) Viewed 10894 times

center Bookeye4 KIC 600dpi mode.jpg (117.67 KiB) Viewed 10894 times

Those are the 5DSR, T2i, ELPH 160, and the library's Bookeye4. (On preview, I guess you can see the image captions and don't really need me to repeat it.) They are all un-resized crops of the center LPPM target, the best ones, the outer corner LPPM target were all pretty consistently 1 or 2 steps down. Also the Bookeye4 was not lit by the window, it was lighting itself the usual way.

These last 2 are the demonstration of the chromatic aberration.

corner ELPH 160.JPG (74.92 KiB) Viewed 10894 times

I really like how clear the 5DSR is, even with the cheapy $150 lens. Yeah, cheap is relative... That cameras is what, $3,000? I still like the thought of a $6,000 pair of cameras compared to the $16,000-$20,000 Bookeye4 if I ever decided I wanted that much image quality. (It's way beyond the scope of my projects, but I can always dream.)

Sorry, I can't seem to find the SD cards that had the 6D or A2200 test images. The A2200 was pretty much what you would expect compared to the ELPH 160 - 20% less pixels and about 1 or 2 steps less LPPM. The D6 was not impressive until we started trying high ISO. Normally the noise would ruin it, but under 800 the noise was almost non-existent. The LPPM was 1 step better, 2 if you stretch your definition of "clear", and compared to the T2i I wouldn't spend the money for the quality improvement just for book scanning. The 5DSR seems to be over-processing edge detection, and there is probably an option somewhere in the camera to fix that, but my friend just got his new toy and we couldn't find it in the menus during the time available. Again, sorry that you have to take my word on the other 2 cameras until I can sort out where I put the SD cards.

[dnwst]

Thanks so much for the comments all, very helpful and I'm very grateful that you took the time to reply! I decided to go with the A2600 because of the currently better support. I realise there are much better cameras out there, but these should certainly be good enough to get me started. I might just be able to get DSLRs depending on how successful my crowdfunding campaign is, but that's an entirely different discussion.