Jump to content

Experimental Camera at Stanford


jonk

Recommended Posts

<p>Cool! This was really just a matter of time. HDR methods are gaining in popularity, for better or worse. Stitching software has been around for a while, and there is software that allows one to augment depth of field with bracketed focus in the same way that HDR images are assembled from bracketed exposure. It's even possible to eliminate moving objects (e.g. cars and people) from images, leaving only the stationary objects (e.g. buildings). Given that all this is happening, it's only a matter of time before it's incorporated into our cameras.</p>

<p>John, Nokia, Google, and Apple are already onboard. Looking into my foggy crystal ball, Madam Sarah predicts that the cell phone camera will never be allowed to surpass the high-end dSLR in its capabilities. I'm sure Canonikon et al. will manage to keep up with the crappy cell phone, and then some.</p>

 

Link to comment
Share on other sites

<p>Okay, so it's an over-powered digital camera running Linux, with a bunch of automatic post-processing built in for HDR, shake correction, etc., and the ability to modify the software or add your own. This is actually less revolutionary than it seems. There are already third-party firmware add-ons for some digital cameras (Magic Lantern, for example), and there are already cameras with built-in HDR. The real point of this thing has more to do with programming than photography, as the article itself indicates:</p>

 

<blockquote>

<p>"I actually think it's a sneaky way of introducing the concept of programming to people who might never have considered it," says Hanspeter Pfister, a professor of the practice of computer science at Harvard. As photographers who worked with film often became chemists in order to modify their art, so too photographers who work with digital cameras could become programmers, he says.</p>

</blockquote>

Link to comment
Share on other sites

<p>Michael: You can trust Stanford intellectual property managers that hey make as much money out if it as possible. The profit these days is in the intellectual property, not the manufacturing. Nobody just hands off these things unless they have no commercial value.</p>
Link to comment
Share on other sites

<p>The further into the article I read, the more I felt that the good professor maybe wasn't very current on what the most recent digital cameras already do.<br /> The examples look considerably less capable than a lot of the HD pictures we see in many portfolios here. Even with a one-picture image, you could get a long way there with just highlight and shadow controls.</p>

<p>If you plug your digital camera directly into a portable computer and carry the whole thing with you, you'd do better than this Stanford "one-shot" as they used to call them in the early color days.</p>

<p>Is it totally irrelevant that Leland Stanford Junior University was one of the key players in that marvel of computer science, the Strategic Defense Initiative? The Stanford Research Institute also has widely backed up Shroud of Turin, Face on Mars, Remote-viewing and other cutting edge projects. %}</p>

Link to comment
Share on other sites

<blockquote>

<p>If you plug your digital camera directly into a portable computer and carry the whole thing with you, you'd do better than this Stanford "one-shot" as they used to call them in the early color days.</p>

</blockquote>

<p>How would carrying your laptop would allow you to do three exposure blending in camera with your current camera?</p>

Link to comment
Share on other sites

If they want to really impress people and make some real money, they should develop a way to make an HDR look natural

instead of as though it were taken on another planet. The HDR concept is great except they tend to look like the pages of

a graphic novel or a Dick Tracy comic strip.

Link to comment
Share on other sites

<p>I forget where I saw it, but I want to say that about six months ago or so I saw an article on an "open source" digital camera at MIT. The camera body in that article looked a lot like the one in this article. Could it be that smart people across the country are all working on the same or same type of project? It seems plausible to me.</p>

<p>I think it's the same project.<br /> http://graphics.stanford.edu/projects/camera-2.0/</p>

<p>I think it'll be ten years, but when we get control of the software more, and won't have to rely as much on proprietary methods, then we'll see what's up. By the time we get it, of course, every cent of profit will have been milked out of the thing.</p>

Link to comment
Share on other sites

<p>Dan, HDR as currently implemented (with tone mapping to cram the dynamic range into a lower dynamic range format displayable on a monitor or printable on paper) will never, ever look natural. Never. That said, an HDR (and non-tone-mapped) image file, with many, many stops of dynamic range, would be fantastic. It could be fractured into multiple layers of different contrasts and judiciously cut and blended. I'd love a camera that could do that for me! OTOH, a sensor with deeper wells (maybe a higher voltage sensor?) might do the same thing.</p>
Link to comment
Share on other sites

<blockquote>

<p>The difference between having just a camera and a camera with laptop connected to it is huge.</p>

</blockquote>

<p>Depending on the laptop, maybe in some dimension like total mass, but is this really a <em>technological breakthrough</em>?<br>

Another breakthrough, let's make everybody learn LINUX to use their camera ;)</p>

<p>What an improvement over the current situation. LOL</p>

Link to comment
Share on other sites

<p>Seems like the industry jumped out way ahead of the academic by about 6 years. I think the next step would be a stereoscope type digicam. Two sensors would work better for in camera HDR and 3-D type applications. And if you had them alternately firing you could get twice the fps for slow-motion stuff. Throw in a third lens for IR and and a Lomo filter and voila!</p>
Link to comment
Share on other sites

<p>I don't get the idea that HDR can't look natural. Obviously there are a million over the top examples here on PN, which I personally dislike rather strongly, but plenty of people are doing natural looking exposure blending. I do it all the time with blown highlights, by reprocessing the same raw file or by using a bracketed exposure, then selecting highlight areas and copy and pasting from one photo to another. The results are very natural looking.<br /><br /> 1. Reprocess raw with exposure reduction and output, or use bracketed exp.<br />2. Select>color range>highlights<br />3. Expand or contract selection as needed ... feather<br /> 4. Copy>paste into photo with blown highlights>align<br /> <br /> My goal is to make something natural that matches the scene, not to create a frankenphoto. It's not hard to do and it's effective. It would not be hard to put the same process into the camera.</p>
Link to comment
Share on other sites

Brett and Sarah,

 

Good comments from both of you. Thanks. I was being a bit facetious, perhaps because I had a certain expectation for

hdr, i.e. that it would look natural if you didn't process it too far. Unfortunately, the opposite is true. You would have to work

very hard to make it look even remotely natural, but what pops out of the software automatically looks really strange. When someone

comes up with an hdr tool that leans more toward the natural look, I think it will be a big hit and I would definitely give it a

try. But I have given up on the whole idea for the time being.

Link to comment
Share on other sites

<p>Brett, that's a good approach to taming blown highlights. However, that's hardly the conventional HDR approach. What makes HDR work look unnatural is the tone mapping step, which you and I don't do. I think your approach might be replicated by starting with a non-tone-mapped HDR file and applying a sigmoidal contrast curve that would compress the highlights (and also the shadows). As I said, I have nothing against HDR, per se, but it's the tone mapping that makes me ill. It's the tone mapping that has become conflated with the term "HDR" and yields that "HDR look."</p>

<p>Dan, I admit there's subtle HDR/tone mapping, but it only yields subtle improvement in dynamic range, and it still looks like "HDR" to my eyes. My biggest gripe about tone mapping is what happens around a high contrast border. The already-high contrast of that border is accentuated. The highlight details on the bright side of the border become washed out (in the halo), and the shadow details on the dark side of the border are crushed. Using my general approach of combining layers, I often make a hard (but feathered) "cut" along a high contrast border, e.g. land and sky, and blend different contrast layers on each side of the "cut." I don't think an automated algorithm will ever do that, and it's not something that can be achieved by sliding sliders.</p>

<p>BTW, maybe you have it all wrong, Dan.  I think a "comic book" mode would be a big seller!  Just look at all these "cartoon yourself" ads we see on PN!  ;-)</p>

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...