See the URL examples above.   Basically dont nest product pages under sub categories or category folders.

You should separate your paginated product list pages from your product pages (as far as the URL structure) or only use main categories in the URLs for products.

Consider these example product list pages.  You have your main list of products

domain.com/category1/sub1/

Then you have pages 2-n that lets people browse products

domain.com/category1/sub1/page2-n

You could have other examples of this (your example would be)

domain.com/category1/sub2/

domain.com/category1/sub2/page2-n

In your example, you have a product that can go in sub1 or sub2 and so you dont want

domain.com/category1/sub1/product-abc

domain.com/category1/sub2/product-abc

as yes - these would be duplicates and cause issues.  Yes you could use the canonical to link to the main product, but what would work better is this

domain.com/category1/product-abc

If you get rid of the subcategory in the URLs you can take advantage of having the KW from the main category etc.  Likewise, if you had a product that could sit in multiple main categories, then it may be you need to use a the approach of

domain.com/products/products-abc

and just completely separate your product URLs from your category URLs.

It would seem that with the use of photographs you will need to always have multiple categories and so this may be your best approach.

You need to sit down with a big whiteboard and map all of this out as it will get complicated real quick.  Then run all your products through it and see if it makes sense.

If you read the article I mentioned above, it really does not hide anything from Google, but you are clearly saying that these links are of no value.

I would use the 301 redirect.

In your case, yes that would save a step to just put /ads/ into robots.

Google will follow a 301 and will transfer almost all the link equity/or lack of it.   Common practice is to 301 redirect through a page that is in robots.txt.  This prevents Google from following while moving the user along.

More here http://makeitrank.com/affiliate-links-seo

It may take a while when the pages you are deindexing are not crawled as often by Google.  You just have to sit back and wait a bit.

Two other points.

Look in your Analytics.  If you delete all those pages, how much traffic do they bring in to start with?  If it is only 5% of traffic, then expect to lose that much.

One correction on the use of robots.txt vs the meta tag.  Robot.txt stops Google from crawling, but will not remove pages from SERPs.  Noindex meta tags on page will get them removed.  Use the former and you will be happier.

Thanks DP for the input!

This is also advised as a part of the on page optimization within the Moz grading system

"Although the canonical URL tag is generally thought of as a way to solve duplicate content problems, it can be extremely wise to use it on every (unique) page of a site to help prevent any query strings, session IDs, scraped versions, licensing deals or future developments to potentially create a secondary version and pull link juice or other metrics away from the original. We believe the canonical URL tag is a best practice to help prevent future problems, even if nothing is specifically duplicate/problematic today."

Well, you can move to a black or gray hat technique of link building, get results (maybe), and if you do, just have them later backfire on you when Google updates the algo and puts you back to square one.  

I would suggest continued education such as http://moz.com/academy or other related sites. There is a ton of info out there, just have to see what works for you.

Probably not.  If you are going to pay $500 and get a link wheel, you will end up with a bunch of spammy/low quality sites linking to each other and to you and that will just kill you.   In theory, if all the sites were of high quality and were on related topics, sure it would help, but the time and effort and cost would not be worth it.

I would agree that use of the canonical tag is great, I would not say that it is the most optimal solution in this case as you have paginated results

http://searchengineland.com/the-latest-greatest-on-seo-pagination-114284

http://googlewebmastercentral.blogspot.com/2012/03/video-about-pagination-with-relnext-and.html

The use of rel next prev would be more appropriate in that case.  It has the advantage of also letting the link juice flow properly and it is what Google "expects" to see.

Now, if you wanted to be more conservative with this approach, you could add the meta noindex so that you also get all the other paginated pages out of the index, but this is an optional step.

One other thing to think about, if this is not a pagination issue, but this is more like a search result or resort of the same page, I would no follow links to those pages and noindex the resulting duplicates.  You have to think about crawl efficiency and if you are having Google crawl a bunch of thin pages that you are trying to canonical to a parent page, you are wasting Google's time.   Google will only spend so much time on a site spidering.   Do you want it to waste time spidering a ton of pages that dont matter?  Sure, the canonical would give Google all the right signals of what page goes where, but why would you want it to waste time doing that.  You would rather Google spend time on your most important pages and spidering and reindexing those.  Think about it, if you are going to ask a math savant to help you with your homework, are you going to have him/her spend time helping you with 1000 simple addition problems?  No!  You would go right to the more important/complex items.

http://searchengineland.com/how-i-think-crawl-budget-works-sort-of-59768

Anyway, hope this helps give you another perspective.  Someone will probably say, well, this only matters on larger sites etc.  I say no, it matters on all sites as you always want to have your best foot forward when the spiders come a crawling.