Tuesday, August 9, 2011

Where the anti-JPG bias came from...

Also in this issue:
  • Stories from New Scotland
  • Manual Exposure for Alpha 33 / Alpha 55's Video 
  • Ebook for the A560 / A580 now available in Spanish! 
Where the Anti-JPG Bias Came From


I don't want to add to the RAW vs. JPG religious debate.  But I thought it would be useful to reveal just where the huge industry-wide anti-JPG bias came from.  Have a look at this picture I took in 2001 using a very early digital camera:

This picture was taken with a Minolta RD-3000 DSLR, a pioneering 3-megapixel camera (and you can learn more about its unusual architecture here.) 

Click on the picture and make it as large as possible.  Or if you're reading this via a newsreader, let me take the liberty of showing you a crop at 100%:

Can't see it the problem yet?  Let me do something unreasonable, then: I'll blow it up to 300%.  (Click to make it bigger).

Yeow!  At 300%, you'll see two things: 1) pixelation, and 2) pretty horrific .jpg compression artifacts, especially around the teeth and beard areas.  And these early .jpgs had two things working against them: 1) the sensors were smaller, so there were fewer megapixels to work with, and 2) because memory cards were small and expensive, and consumers wanted a lot of vacation shots to fit in the camera, there was a lot of incentive for the engineers to make the .jpgs take up as little room on the memory card as possible.  This meant heavy compression, and this resulted in exceptionally strong .jpg compression artifacts like you see above.

This, ladies and gentlemen, is where the "JPGs are inherently inferior and so I'll never shoot in that mode - EVER!" mindset came from.  And back then that sentiment was quite justified.  But today memory is cheap and plentiful so the pressure to heavily compress the JPGs has eased.  How do today's JPGs compare?

Have a look at a shot I took of Grandson #2, using an A900 set to medium-quality JPG.  (Again, click on the image to make it bigger). 


Now lets give this image the same treatment: 300% on the right eye (click to make it bigger):


Here I can only see pixelation - if there are any .jpg compression artifacts there, they're incredibly hard to see.  And this was a MEDIUM-quality .JPG!  (Anyway, this kind of pixel-peeping is not a meaningful way to evaluate image quality.  If you have to zoom in to this degree to see anything then the image quality must be pretty good!)

So today's DSLR JPGs are of higher quality than most people realize.  Are they flawless?  Of course not.  I'll bet I can design a test that will make the compression artifacts clearly visible.  I'll shoot a high-frequency subject ("high-frequency" meaning "Lots of bright white and dark black right next to each other") - exactly the kind of situation which breeds such artifacts.  Here it is: A picture of an old textbook (if you must know it's the Electrical Engineers' Handbook by Pender and McIlwain, circa 1936, which belonged to my father):


Requisite 300% zoom-in below (click to enlarge):


Hmmm... it looks like pure pixelation!  The compression artifacts here are very hard to see.  Wasn't expecting that.  Let me see if I can dig up an older example...  Aha!  Here it is, from Sony's first DSLR, the A100, taken in 2007:


And the 300% crop (click to enlarge):


If you look very carefully at the edges of the black letters you'll see very small white box-like structures.  There it is!  I must say I'm surprised at how difficult it was to show on a modern camera like the Alpha 55.  (Just as it's difficult to show the internal ghosting on the A55, even if you design a test specifically to reveal it!) 

Conclusion: While JPG's bad reputation had once been well-earned, modern .jpgs are of such high quality that if my light is good I tend to shoot only JPG.  I'll switch to RAW (or RAW + JPG) mode when:
  1. I'm not sure of the white balance or the exposure,
  2. I'm using very high ISO (whose noise can be removed more effectively using Lightroom), or 
  3. the light is bad / the dynamic range of the image is a little wide. 
Notice that "whenever I have a persnickity customer who insists that I submit the images in .TIFF format" is not on that list.  Whenever I have a customer like that, I just convert the .jpg to a .tiff and send it off.  None of those customers ever noticed a quality problem - they just blindly spouted out conventional wisdom from 10 years ago (and didn't read this blog). 

(By the way, if you want to see a great example of compression artifacts, most cable TV companies employ LOTS of compression to squeeze more channels through that skinny little cable.  You can see it changing constantly in dark backgrounds.  You'll never look at cable TV the same way again!)

Stories from New Scotland

“I remember a time when the poor kids at school would try to trade their lobster sandwich for a peanut butter and jelly sandwich at lunchtime”.  So says a Nova Scotian who grew up in the North Atlantic region, where Lobsters were considered to be “garbage food” due to their status as bottom-feeders.  Prisoners dined on Lobsters a century ago.  Today, of course, it’s considered a delicacy (and McDonald’s restaurants up here even serve McLobster sandwiches.)  How fickle and arbitrary our tastes are!

I heard this last week while conducting my most recent Seminar and Field Workshop in Nova Scotia, Canada.  This actually marks our third seminar in this region - previous events have proven so popular that we keep getting asked to return.  

Proof that you can still get great shots when the weather isn't great (like this shot which was taken in the pouring rain!)

Did attendees have a great time and remark about how my approach to teaching photography is refreshingly different than anyone else?  Let me share some quotes with you from the after-the-seminar questionnaire: 

“Unpretentious photographer sharing his knowledge and experience shamelessly!   Gary, you were AWESOME!”  - Lisa Buchanan

“This is the best photography class I have taken to date.  Thanks, Gary!” – Heather Cromwell

“Stop going to forums and run to Gary.  I used to be a lensaholic but Gary helped cure me of my pixel peeping addition.” – Jim St. Croix 

And this month I'll share just one picture with you that was taken by Field Workshop attendee Robert d'Eon.  We were all at a lighthouse shooting at sunset, when suddenly...
Robert explains... "At first I was getting frustrated, since Brenda was blocking the shot I had pre-visualized.  Then I said "what the heck" and just took it." :-) 

I started the Friedman Archives Seminars because I was tired of hearing the conventional online wisdom that says "in order to take good pictures, you have to shoot in aperture priority mode / have to have $1000 lenses / have to shoot RAW".  The problem with this advice is that by itself it won't improve a beginner's photography.  (At all.)  There are secrets to taking photos that can make a stranger say, "Wow!", and you can employ these techniques by using a lowly point-and-shoot (as I demonstrate on the FriedmanArchives.com website, where about 1/25th of the images there were indeed taken with a point-and-shoot.)  So I've been sharing these secrets in a series of fun, interactive seminars all over the world, which have been enthusiastically received.  And we'd love to come to the city that's close to YOU!

The last two seminars of 2011 will be in Ottawa and Nashville, then Copenhagen and London in 2012, and then....  well, after that these highly-acclaimed seminars are going to change.  Stay tuned for more on that.
Ottawa, Canada September 3-5, 2011 (Two days lecture + one day field workshop) Learn More and Sign up!
Nashville, TN October 1-2, 2011 Learn More and Sign up!
Copenhagen Seminar April 21-22, 2012
(There will also be a Field Workshop and a separate lecture on my days as a NASA engineer)
Learn more about Day 1
Learn more about Day 2

Learn more about the Refresher Course

Learn more about the Field Workshop

Learn more about the NASA lecture
London September, 2012 Click here to register interest

Manual Exposure for Alpha 33 / Alpha 55's Video 

Many folks online have derided the A33 and A55's video mode since the perception is that there's no control over the exposure variables (f/stop, shutter speed, ISO, etc.)  Most of these limitations stemmed from the fact that the phase-detect autofocus needs to have the f/stop as wide open as possible to work right.

But what if you didn’t need autofocus?  What if you switched your AF off and just focused manually, like the Hollywood cinematographers do?  Well, then you can specify your f/stop after all!  Here’s what you have to change:
  1. Set the camera to Manual Focus (the switch either on the base of the lensmount or on the lens itself).
  2. Focus on where your subject will be.  You can optionally invoke the Focus Magnifier function to do this.
  3. Set your camera to “A” (Aperture Priority) mode and choose your f/stop via the front control dial.
  4. Record away!  You will hear and see the f/stop of the lens close down when recording mode starts.
When your camera starts to record video, it will also use the currently set white balance, creative styles, exposure lock, exposure compensation, AF area (if you’re autofocusing), and metering mode.  So you still do have quite a bit of latitude to get your footage to look the way you want.

 I used the A55 to create an "artistic" (ha!) underexposed product video for a digital guitar I'm thinking about importing.  Because I wanted control over how the scene was exposed, I did the following to give it a look I never could have achieved using "auto":  First I metered for the black backdrop and hit AEL (toggle) to lock it in.  Then I set Exposure Compensation to -2 to make the background look black instead of grey.  Then I hit the record button and shot away!  The exposure would not change after that, even if a person wearing a white shirt would come into the frame.  It looked quite professional.  You can see it below:


If that all sounds a lot of trouble to go to, then you'll love the latest rumors that I read on another photo blog I neglected to mention in last month's post: www.sonyalpharumors.com.  Sony is expected to announce the A77 (among many other products) at the end of August, and among its very impressive specs (24 megapixel APS-C sensor, 12 fps shooting speed, 60p AVCHD 2.0-compliant video recording), it is said that you'll have complete manual control over all the shooting variables when shooting video.  This should make many of the professional videographers out there (well, the ones who have time to hang out on online forums, anyway) very, very happy. :-)

Ebook for the A560 / A580 now available in Spanish

My translator, Francesc Garcia, has been working overtime.  In addition to working on my updated guide to the A33/A55 (which includes firmware 2.0 AND the new Alpha 35), he also has just finished translating the A560 / A580 ebook, which you can read more about here:

http://friedmanarchives.com/alpha580/sp-index1.htm

(Yeah, I know, the links for the printed books go to the English version.  That should be fixed by tomorrow.)

Until next time,
Yours Truly, Gary Friedman

Handheld Panorama stitched together by the Sony A55 camera.  Notice what can happen if you don't keep your camera straight - a jagged ocean horizon.  Let this be a lesson to you.

==============================
Check out the last two seminars of 2011 - Ottawa and Nashville!  www.FriedmanArchives.com/seminars

6 comments:

  1. I don't consider myself in a war with .jpg. I use it on a daily basis when I post images to the web or as enclosures in email. It certainly has its place as a file format, however I strongly believe that that place is now very much, strictly a post processing export format, and should rarely, if ever be used for export directly from a camera. This has nothing whatsoever to do with a .jpeg image's quality, or lack thereof.

    When you choose .jpeg as the export file format from your camera, you are, forever choosing to accept the camera manufacturer's choices for initial manipulation of the image. This has little to do with destructive compression, although that can be a problem too.

    As Gary said, much water has run under the bridge since cameras started exporting .jpg. Indeed, cameras have become much more sophisticated and the images they export have become much better, BUT output devices such as monitors and printers have also matured. What has NOT matured is the .jpg standard, which is fixed in stone.

    Today's cameras shoot in 10, 12, 14 and even 16-bit bit depth, giving dramatically increased dynamic range. IPS monitors and pro-level printers support these higher bit rates and also support wide colour gamuts… far wider than sRGB or even AdobeRGB. By choosing the .jpg file format, as well as destructive compression, you automatically choose a limited 8-bit, sRGB output that permanently and visibly limits the ultimate quality potential of an image… regardless of enlargement size.

    By shooting RAW, working in a non-destructive image editor and converting to another file format later you avoid all these pitfalls and more. The original RAW converters were good but not great. There have been considerable advances in the quality of RAW converters. Because I elected to shoot RAW only, I can re-process all of my images using these new RAW converters and take advantage of the improvements. I will be able to continue doing so each and every time a better RAW converter emerges. I'm not locked in. Likewise, because RAW output does not convert bit-rate or colour space, I can take advantage of continued improvement in output devices, although I admit that we already have devices with better colour gamut than AdobeRGB. We need to convince camera manufacturers to support ProPhotoRGB in-camera.

    I completely agree with Gary when he says memory is cheap. I also agree that it can become confusing if you are constantly having to remember to change file formats in your camera. That's why I set my camera to RAW and the AdobeRBG colour space when I bought it. The settings have never changed. Everything gets shot RAW/AdobeRGB. I convert to whatever file format I require in post.

    Simply put, shooting RAW gives me choice… both now and in the future. .jpg does not.

    Every photograph is an instant in time, recorded. You only get once chance. Why needlessly limit your choices?


    Gerry Curry
    Gerry Curry Fine Art Photography
    and Printing

    ReplyDelete
  2. Again an interesting and informative newsletter presented in an entertaining way. Keep it up. And Thank you!!!
    YOur friend S.

    ReplyDelete
  3. when you export as JPEG you:
    - Lose the ability to correct whitepoint
    - Reduce the ability to correct under/overexposure
    - Lose the ability to use the image in HDR
    - Reduce the ability to use many third party applications such as Lightroom.

    When you export as RAW you:
    - Reduce the speed of your camera in continious mode (at least, you do once the camera's local buffer is full).
    - Reduce the number of images your camera can store.

    MAin reason I shoot RAW is because I believe the ability to tweak an image after the event using Lightroom and Photomatix et all more than compensates for the speed advantages of JPEG.

    ReplyDelete
  4. Hey, I shoot RAW too! But I've found that if my light is good (and the exposure is right for that light), then my shots don't benefit greatly from RAW. If I'm unsure about anything, I'll shoot RAW and figure it out later. :-)

    ReplyDelete
  5. Gary's last comment touches on the simple reason why I shoot only RAW. Having such complete control over a scene that I know I will not have to coax detail from the extreme tones of the image (where JPEG doesn't give you many bits to manipulate) happens... well, almost never.

    It's a valid approach to try to make the image coming straight out of the camera require little correction, but my approach has tended to be to think of the camera as the device that captures the maximum possible information from the scene, knowing that later steps in the workflow will turn that information into the most pleasing image. That approach requires RAW (14 bits per pixel is 64 times as many tone levels as JPEG's 8 bits), and keeping as much of the image high up on the histogram as possible. Highlights can even be overexposed, as Lightroom's Recovery feature brings detail back if slightly overexposed. With my A55 I usually expose +0.3 or +0.7 over for this reason.

    The resulting image doesn't always look like what you want to publish (it's usually a bit flat and too bright) but if you consider the goal of the image capture phase of the workflow as capturing the maximum amount of information from the scene, that's how to do it. The following workflow steps can then accentuate detail in high or low brightness areas as you wish. WIth this approach, that detail is there. With a typical JPEG, unless it happens to be *perfectly* exposed, it won't be.

    ReplyDelete
  6. Though jaggies and artifacts are on some level intrinsic to JPG compression algorithms, their disruptive visibility is more an indictment of unartful photographer's mismanagement of their resources than they are an indictment of the JPG standard itself -- a standard that when used appropriately works fine for knowledgeable folks.

    Jaggies are usually the results of up-sizing COPIES for publication from ANY file, not a file format problem, and if anyone has problems with jaggies during ORIGINAL capture and archive, they have a bigger problem than merely understanding JPG.

    Non-image artifacts relate to compression and manipulation where someone tries to compress an image below the level of information they wish to convey, or tries to expand gamma beyond the bit resources in an image -- "tweaking" COPIES rather than returning to the OIGINAL to make new adjustments.

    Because JPGs are 8-bit, and compressed (in blocks of 4, 8, and 16 pixel groups), and often inappropriately resized, but more importantly, because they are the 99.44% imaging distribution standard, JPGs will always appear to be the prime source of jaggies and artifacts.

    But, it's not JPGs that's the problem, it's person playing with the JPGs that's the problem.

    JPG COPIES are an output medium, not intended for further manipulation of any sort (though resizing DOWN may be acceptable).

    Raw ORIGINALS are a capture and archive medium offering maximum resources for any subsequent tweaking and manipulation (then saving as a distribution JPG sub-set COPY perhaps).

    Raw and JPG both open quickly and painlessly in free http://www.picasa.com/ and free http://www.irfanview.com/ and many other free, and costly, simple, and sophisticated programs.

    Large, fast storage cards for cameras are CHEAP.

    So, there's hardly any reason to be afraid of Raw, or JPG for that matter, once you understand what each is good for.

    Raw for ORIGINAL capture and archive;
    JPG for distribution COPIES.

    Click!
    Love and hugs,
    Peter Blaise

    ReplyDelete