Extended Dynamic Range - How many stops?

I am starting to doubt SMaL's claim of 120db extended dynamic range!

I left my basement and took 25 pictures on both a micron imager and

a SMaL imager. The lens in SMaL camera is 10mm in diameter and lens

in Micron camera is 14mm in diameter (should favor Micron). I

alternated between cameras with less than a minute between similar

shots. Neither camera has any ability to override automatic

electronic rolling shutter.<P>

The comparison below is one I thought demonstrated maximum use of

SMaL's patented "advantage". Does anyone see more than 4 stops of

difference? That is my best uneducated guess. I still think that by

removing bayer and using a pinhole lens, I might be able to double

range but doubt that anything near 20 stops is possible.<div></div>

The top photograph is totally underexposed. How can you make any kind of comparison between two photographs taken with entirely different (and in the top case, incorrect) exposure settings? All you've proven this time is that whatever camera took that top picture has a lousy/broken metering program.

 

The bottom picture displays worse quality than my old plastic Canon Powershot 350 (circa 1997) might have taken, and certainly no better dynamic range. On this basis, I think you're absolutely right: SMaL amounts to nothing. Maybe now we'll hear the end of it from you?

Just finished reviewing several months of your posts and I'm convinced that this SMaL crap you keep rambling on about is nothing more than trollsh*t. You've even hijacked at least one other thread by your responses to drag SMaL into it. You haven't gotten one single hint from other posters that YOU seem to be the only one who gives a rat's patootie about SMaL. Maybe we need to take up a collection for you to buy a clue? Everyone else is busy with their photography while you seem to want to tilt at the SMaL windmill . . .

What about BiG? That would be better, right?

I recently tested the SMaL imager versus the consumer grade Panasonic FZ-15 in a shot of Air Force Test charts to determine if the 120db of dynamic range could result in improved contrast of fine details.

 

Here's a full sized crop from the FZ-15...<div></div>

Here is the equivalent shot taken with the SMaL imager. Do you think that I may be having exposure problems with the patented SMaL imager?<div></div>

What's truley surprising is that shot from the FZ15. ~1500 lpm if this was done like they do it a DPReview.com. But I agree, you have a problem with the SMaL imager. A bit of underexposure, I suspect. I understand they are difficult devices to work with. Has anyone ever gotten a really good shot with one? Never have like mine.

Steve, I'm fairly certain that you need to peel off your Bayer filter. The instructions printed on the outside of the cardboard box the camera comes in actually tell you to do this, but many people forget to read them. The Bayer filter is to the sensor what a lens cap is to a lens, and is supposed to be removed before taking pictures.

When is this troll's account going to be suspended?

I suspect SMaL exists get someone to acquire them. (Which I see Cypress did.) A strategy somewhat along the lines of: 1) invent product with ridiculous claims, 2) embellish with marketing-speak, 3) give product away or pay someone (like Radio Shack) to use it for market share, 4) get ISO 9001 certified, 5) get acquired quickly before running out of cash!

Jesper, naw, let's keep him around - it's light-hearted humour, and we can all do with a good dose of that from time to time to bring us down to earth (including me).

Sorry to create such a disturbance; but I am looking for a true evaluation of comparable imagers. The Micron camera does not have a shutter and the SMaL camera's shutter is open whenever camera is powered up - both rely on an algoritmically controlled electronic rolling shutter determined by imager electronics. Both produce similar shots of scenes with only a few stops of dynamic range. Perhaps the Micron underexposed because blue sky at top only impacts a quarter of the pixles in top rows. Maybe the IR filters work differently - oddly direct Sun shots show with a blue center on Micron and a pink center on SMaL and only a single stop difference (my estimation). I would still appreciate comments as to how many stops Micron underexposed and an estimate of how many stops the SMaL photo covers. Can we keep to strictly technical evaluations? I believe any method of extending dynamic should include pixel color as part of the algorithm (guessing that SMaL does not is why I want to remove bayer). And if Joseph is still around I would enjoy meeting on a sunny day and taking some simultaneous wide dynamic range outdoor shots. In addition the high quality Air Force test shot above looks to me to cover at most 4 stops.

Yes, but what is the point of having technical discussions about something that amounts to a cereal box prize? The image quality is so poor that it's pointless to talk about the technicalities of it - if this is honestly the best this technology can come up with, then it simply isn't worth pursuing at all - the issue is moot. I've looked through your gallery of SMaL samples, and every single one of them is atrocious on a purely technical level - I am not even bringing aesthetics or composition into it - they are just technically horrible images. My newspaper routinely prints better quality images on their front page.

 

Our latest box of Special K contained a bright red plastic whistle that emits a single high-pitched squeek (possibly in that magic range of 120db). Do you think I should find a forum for musicians to ask experts to discuss the technical merits of my plastic whistle? Should I spend 18 months in my basement (sorry, it had to come up sometime), experimentally drilling holes into similar plastic whistles to try and turn it into a flute with 4-octave range?

How about the Captain Crunch whisle that was at 2600 cycles; and could be use to grab the Ma Bell long lines 1-800 trunk line ?

Kelly, hey - I remember that whistle! It was big news at the time, wasn't it? I don't think my red whistle does anything except irritate people...

Kai, the point is that some folks are actually interested in immature technology that might have future relevance.

 

Did you get nauseated when you first heard of digital imaging or did you really find pixelated, mushy 480x640 images to bode well for the future of photography?

 

Or does it just make you feel hip to disdain what you're not interested in? I mean, if so, that's cool because I don't like hardcore hip-hop no matter how many Grammys are awarded to mumbling thugs.

 

Not only are digital cameras with greater dynamic range coming, some of them will feature adjustable dynamic range and "smart" dynamic range that can read a scene and adjust for it, rather like Nikon claims its 3D Color Matrix Metering can do. That will beat the socks off the current need to blend two or three identically framed photos taken at different exposures in order to preserve highlights and shadow detail.

 

Eventually we'll also see "smart" automobile windshields that can adjust selectively for hot spots from sunlight, oncoming headlights and glare, based on the driver's position.

 

But you probably won't like it.

Lex, I don't understand how what you say relates to Ray's images from plastic SMaL cameras. I don't object to progress - but this is not progress at all. As I said above, the images coming out of these plastic SMaL cameras are inferior to the plastic camera I had nearly 10 years ago. Where is the progress in that? I mean, if extended dynamic range cameras are coming, then of course I'd love that as much as anyone else, but you aren't seriously suggesting that this SMaL thing is evidence of the immenent arrival of such cameras, are you? If you are, then you're seeing something most people aren't...

<i>I would still appreciate comments as to how many stops Micron underexposed and an estimate of how many stops the SMaL photo covers</i>

<br><p>

Generously, 6-7 stops.

I see potential in a concept. A $10 plastic camera that doesn't hit the mark hasn't invalidated the concept, only the execution.

 

If you read Ray's bio you may get a better understanding of what drives him to experiment with this stuff.

Hi Lex -

It's not that I'm discarding the concept or its potential (you appear to have derived that from my post(s) erroneously). I am discounting the relevance of trying to prove - or for that matter, disprove - the concept by using abysimal images and methods of comparison that offer no way of comparing anything at all. If we're supposed to take this seriously, the least we could do is be a little scientific about it.

 

Based on all the SMaL images we've seen (here, and on SMaL's website itself), there is not a shred of evidence that the SMaL implementation has any potential outside of the security camera systems it was apparently designed for. I'd be the first to applaud them if they came up with something usable, but perhaps someone clever like Ray will beat them to the punch by coming up with an alternative that really does have legs.

 

Cheers,

Kai

You can't make a silk purse out of a sow's ear, nor can you get good images from a crap camera with a crap sensor.

<p>

I don't understand the obesssion. It's not like anything that <em>you</em> can do will improve the technology and the images are crap right now, presumably due to the miserable quality of the sensor.

<p>

It's your time to waste though and we all need something to do. I'd still like to see what happens when you take the Bayer filter off using a belt sander.

<p>

Attached is the result of <em>my</em> "waste of time" project. Conversion of a Coolpix 900 to infrared imaging. A slightly better use than a doorstop or paperweight. Only slightly better though.<div></div>

Ray,

 

I realize that you have a sincere interest in this, but your experiments are so dogged, and yet poorly conceived that taking a light poke at you was almost irresistible.

 

If I'm doing my math right, 120db is 12 orders of magnitude on a log10 scale, which results in something around 40 orders of magnitude on a log2 scale, or around 40 f-stops.

 

Take a look at this EV table to get an idea of what sort of range that provides and whether it is really, really necessary:

 

http://www.fredparker.com/ultexp1.htm

 

From bright, bright light in snow or desert (ev 16) to a night time campfire lit scene (ev 5) is only 12 stops (inclusive). In a scene likely to be photographed for artistic or personal purposes, getting 10 stops is probably more than adequate (Ansel Adams only had something like 10 or 11 zones, right?)

 

If you look at the results for the Fuji S3Pro on this page: http://www.imatest.com/docs/q13.html you will see that it seems to get 11 stops of range.

 

Do we really need 42 stops? Probably not. Especially if the sensor is compressing the 42 stops of dynamic range into a representation that only has around 8 stops (like a typical jpeg file). You don't know if the original scene had 8 stops, 20 stops or 42 stops of actual range. To me, this is fine for surveillance, but not so hot for normal photography.

 

Is it always necessary to shoot 2+ exposures to get extended dynamic range? No. Just use RAW. I'm appending a photo I took last week with my dinky Fuji E550 - I shot the picture in RAW, and the top photo is the "normal" photo. The bottom photo is a very simple adjustment using Photoshop CS's highlight/shadow tool - it is clumsy, but shows you that there is detail in the shadow areas that can be recovered out of the original 16-bit source. You can often squeeze another 2-3 stops of range out of the same RAW file (and the Fuji is a dinky sensor without the big photosites of the DSLR's).

 

If the Fuji S3Pro really has 11 stops worth of detail in it's RAW files, then it pretty much covers the dynamic range you'd expect to see in natural scenes. There is no need for SMaL - all we need is SuperCCD.<div></div>

Old fart Detroiter's viewpoint of SMal's goals:) ...:<BR><BR>Some of the SMaL sensors goals & customers have to do with Detroit; of the straits;); automotive stuff; motor city; willow run; b24's.mile roads; Sanders Fudge; Vernors; Woodward avenue..... You are NOT shooting a darn POW for photo.net You are viewing out the stern/rear end of your old chevy; a new Boeing 787; the sensor is peeping out the rear bumper; the fan tail; the wing; side; so the neighbors kid or dog is not run over; a trailer of luggage is not jet blasted. There is NO time for a driver; pilot; to screw around with photoshop. The extended range of the sensor's SMal deal is many times aimed just for a viewing screen; not a lame digital file to be played with later. Think real time greater dymamic range as the info advert cheerleaders cry.

I think you dropped a factor of 2 (the same mistake I made in the last version of this thread). I believe dynamic range in this context is given by 20 log (I1/I2), so 120db is 10^6, or about 20 stops.

 

The rest of your analysis is still fine though. You don't need 20 stops.

How many stops? That's what I asked. Thanks Jan for only answer to my question. I can laugh at cheap plastic humor (but that is all I can afford for my experimentation). Thanks for all the responses - I find value in each (especially Steve's and Kelly's). Actually if Steve's photo above has 11 stops at sunset, how many more would my sky photo at midday have. Please ignoring plastic quality, can anyone else estimate dynamic between sky and shaded trees in my shot and also Steve's shot? What about Bob's shot? I have lost faith in 120db but would still like to know if range is extended by 1-2 or 4-6 stops! Will a cheap meter be able to obtain a reading of sky and dark shadows underneath a bunch of trees? With my cheap plastic 100 power Radioshack microscope I cannot see bayer - making belt sander removal impossible, but I still might try chemical removal.