Wednesday, July 4, 2018

#NaturePorn Part 2


Also in this issue: Follow up on...
  • Image Search on Local Hard Drive
  • Encryption to resolve the Privacy vs. FBI debate
  • Constraint Theory
  • Chocolate and Bacon
#NaturePorn - Epilogue

Back in March I had started a local debate about how much manipulation is too much.  On the one hand, over-processed images sell.  On the other hand, it's dishonest and can go a little too far on the continuum of snapshots and artwork.

For years I thought I had the right balance, but then a commercial software app made these kinds of over-the-top images embarrassingly easy - even those who have never taken the time to master Photoshop can crank out this level of work.  I could fill my website with crap like this, and possibly license more images.  But I wouldn't be proud of any of it.


This is how the original image looked.
After much thought, I believe I have found the right policy.  Rather than just tell you my answer, let me introduce you to the latest episode of the webseries "Photographers in Cars" where I'm a guest star.  I explain the problem and how Duncan Hines inspired my solution:



(Scholarly note: After we recorded this episode, I did some research and discovered that the Duncan Hines story was a myth.)  But this doesn't change the validity of my conclusions.)  (Scholarly note #2: Be sure to watch all the way to the end, where I show off some ancient photographic museum pieces.)

Image search on local hard drive

Last September I publicly advertised my need to search my local hard drives for a particular image that was similar to a thumbnail I had.  Apparently it's not a common problem and there are precious few programs out there designed to address this specific problem.  And most of what I tried crashed on my 8 TB dataset.  But I did find one that finally worked after working with the tech support team.

Here's what I tried, so you don't have to waste your time re-discovering the wheel:

Imgseek - crashed upon database / directory scan.  Not very bueno.

Visipics - This program merely finds and deletes all duplicates it finds on your hard drive.  It doesn't let you feed it a specific image to find.

Visual Similarity Duplicate Image Finder - after scanning for nearly a week (at high CPU load), it finally crashed at 48%.  (Told it to index only 2 of my 3 drives.)  After contacting technical support, I was told to set the Cache to "No cache" and lo and behold it searched my entire data set in less than a day.

Even more impressive is what I gave it to work with:

The original was a picture of a page in a photo book.

It found the original!
This was with a similarity set to 68%.  This is now my go-to image finding software.  All I need to do is set aside a full day for it to complete its search.

Encryption

Back in 2016 I talked about how there could be a solution to the current ill-informed encryption debate - that you could encrypt private information (on a smartphone, for instance) yet still provide for law enforcement's legitimate needs without having to build a "back door" which could be exploited by bad actors.  Although others have echoed my case (see the bottom of this follow-up blog post from 2017), the proposed solution seems to have fallen on deaf ears.  Even my personal message to the deputy director of the FBI, who has publicly called for a balanced solution like the kind I proposed, has gone unanswered.

This is another instance of "ideas do not catch on by merit alone".  I say this because someone else with more street cred than an overeducated photographer has been evangelizing this very idea, and taking it on the road and presenting it to "all the right people".  You can read much more about him in this article from wired.com.

Another article exposing the same ideas has been published in Techcrunch as well.

Once again, my ideas have been validated.  But what good is being able to solve problems if those solutions fall on deaf ears?

Constraint Theory is Now a Commercial Product

Last September I talked about the milestone of getting the 2nd edition of my Father's book on Constraint Theory out the door, despite him having Alzheimers and didn't remember much about his theory by that time.

A part of the story I didn't tell you was that five years prior, my father received a visit from a "mini-fan club" - 3 engineers from Bosch in Germany who had read his book, and loved the theory so much they started to implement it in software.  My Dad, his graduate student, and I were all impressed.

Three Germans, two Friedmans, and a Graduate student.
Fast forward to today - the software we were shown is now a commercial product, and being sold as a systems engineering tool to auto manufacturers around the world!

Here's a link to a quick youtube video showing you how the software works.




You can learn more about the software (and order a copy for your company!) here.

Unlike most thinkers, my father's ideas have been recognized and acted upon within his lifetime.  I can't think of anything more gratifying.

Parting Shot

NASA scientists have finally combined the world's two best foods - chocolate and bacon - together at last!  I thought I would document this milestone Father's Day gift before taking a first bite.

Until next time...
Yours Truly, Gary Friedman

16 comments:

  1. Re: Image Search
    Gary, since Google among others can do facial recognition wouldn't it be possible to tag a face in a photo and then let facial recognition find all photos with the same face? Of course I presume you'd have to load your entire database to whatever image repository that would be doing the recognition? But in theory wouldn't it at least be possible?

    ReplyDelete
  2. The podcast is really enjoyable!

    ReplyDelete
    Replies
    1. I'll tell Spencer. Please help spread the word - I think a lot of people would benefit from seeing it.

      Delete
  3. Thanks you for the information about Image search on local hard drive. I'm just starting to learn to organize and categorize all my work, and this is a critical part I didn't even think of - knowing that one day I would have to search for duplicates manually.

    Thanks!

    ReplyDelete
  4. I was surprised your fellow passengers were unfamiliar with selenium cell light meters. I had the GE (lost) but still have a Weston (albeit not used in decades tho' it still works)

    ReplyDelete
    Replies
    1. I'm no longer surprised that anyone wouldn't know something. :-)

      Delete
  5. Over-processed images?? Remember, people still buy velvet paintings of Elvis. My dad always used to say; 'there is no accounting for taste'. I agree with the 'truth in advertising' approach you advocate. If people want the over-cooked digital art (rather than a proper photographic piece of art) then so be it. It's their money. Keep up the great work Gary!!

    ReplyDelete
  6. you write "3 engineers from Bosh". Are you sure that you don't mean the Company Bosch from Germany

    ReplyDelete
  7. Still have my Gossen Pilot-2. It uses a selenium cell. It still works great. Which brings up a question. Do selenium cells wearout or go bad?

    ReplyDelete
    Replies
    1. They last between 20-50 years (longer if they're kept in darkness). The needle in mine stopped working so I can't really check it for accuracy.

      Delete
    2. I've had mine more than 17 years. It is kept totally inside leather zippered case and it still works.
      Thanks Gary

      Delete
  8. I just finished Bruce Barnbaum's book: "The Art of Photography A Personal Approach to Artistic Expression." He makes the point that the photographers job is not to just collect photons but to create a composition that expresses what he or she felt about the scene and then process it to produce that result. Much of his work is black and white and much of that is with film. He spends a lot of time speaking about how do develop and bring a negative to produce the image that expresses what the photographer felt that inspired the exposure. He reminds the reader that Ansel Adams famous photo 'Moonrise Over Hernandez' was highly manipulated in the Dark Room with dodging and burning. I suspect he would argue that the image with the artificial sky and reflection of the sky didn't really reflect what drew the photographer to make the image. He does discuss several abstract images he made and also a few that where composites between two negatives. His test seemed to be how well did his image express his feelings. He wasn't driven to make something pretty but to use his tools to express himself. The differentiator in part becomes having something to say.
    I hope I have expressed his ideas fairly accurately. His images are inspiring.

    ReplyDelete
    Replies
    1. Hey, Peter! Bruce's ideas are not new; in fact I often bring these principles up in my seminars. "It's the image you've pre-visualized!!" I grew up shooting editorial style, with images designed to illustrate an idea and therefore had a responsibility to be as truthful as is practical. Art of course is a whole 'nuther world.

      Delete

Thank you for your comment! All comments must be approved by a moderator before they will appear.