Monthly Archives: October 2007

HDR – Another attempt

Sarah and I stayed Saturday night at Elwood Cabin, a Forest Service cabin on Elwood Pass in southern Colorado.  About four miles from the cabin is the ghost town of Summitville, and the site of the Summitville Mine superfund site.  Sunday morning we spent about an hour walking through the ghost town and taking pictures.  Again, I did some handheld bracketed shots, and this time I decided to try out a Linux workflow. The images were combined using qtpfsgui, which is a Linux native GUI for creating and tone-mapping HDR images.



Click the image to go to the Flickr set, which has the original images post RAW conversion (while balance set to Bibble’s “cloudy” preset).  I think the result is not half bad.   I plan on doing a few more images from this shoot. The effect is more naturalistic than my previous attempts.  Unfortunately I think the composition of the image might rob it of too much context, making it a bit uninteresting even if visually pleasing.  I also should have paid more attention to the aperture setting and tried to get the foreground in better focus.  I think there will be stronger images from this shoot once I finish going through them.

Image to image consistency is important preparing HDR images. Your source images must not only line up, they must have identical white balance settings.  You must bracket by shutter speed rather than aperture so the depth of field is consistent.  It’s helpful to shoot RAW for HDR. This way you can override your camera’s automatic white balance and contrast settings. If your camera doesn’t shoot RAW, you can always manually set white balance and contrast settings, but RAW makes it easy to make fine tuning decisions after the fact.

It seems like Linux is thinking about catching up with regards to RAW digital photography workflow. There are high quality RAW conversion apps, and qtpfsgui feels very powerful and compares quite favorably with commercial apps like Photomatix.  The Gimp (open source’s answer to Photoshop) doesn’t quite measure up, mainly since it supports only 8-bit images, but CinePaint, which is The Gimp adapted for film use, has a core that can handle even 32-bit HDR images.  The flip side is that the software is significantly more difficult to learn than commercial equivalents just as Linux demands more from its users than MacOS or Windows. I am not aware of any open-source workflow app at the level of Aperture or Lightroom, but the excellent Bibble Pro from Bibble Labs is available for Linux. The core strength of Open Source software is that as the different pieces evolve and improve, they can be combined into powerful solutions. Given the right momentum around the right projects, it’s possible we’ll be seeing professional grade workflows running on Linux with 100% free software within a couple years. Photography is by its nature a pretty geeky pursuit, so I give it very good odds.

Tags: , , ,

Advertisements

Handmade HDR?

We went out for a day of shooting in September, driving the pickup south over Guanella Pass. The day was blustery and cloudy, which made for some nice shooting conditions. I attempted to shoot several pictures with the idea to combine them into HDR images, however I didn’t have a tripod on hand, so I set my camera for auto-bracketing, set for continuous shooting, and shot off three hand-held shots in quick succession. This produced photos which weren’t suitable for automated HDR creation methods, so I was left to line the photos up by hand and use the magic of layer masks to combine properly exposed areas from two photos. Here’s the result of the first attempt:

hdr1

Clicking on the image will take you to my flickr account, where the picture can be viewed larger. This is from two exposures. The sky is from an underexposure and the foreground is from a metered exposure. The hills in the background are blended between the two exposures. It’s definitely got that faky HDR thing going on. There’s a bit of haloing around the border between the hills and the sky, due imperfections in the layer mask and layer alignment. It’s a kind of interesting, the border provides contrast while maintaining the atmospheric perspective, but it doesn’t feel genuine. For comparison, here’s one of the exposures by itself, processed through Aperture:

20070923-Guanella Pass 304

Again, you can click the image to go to my flickr account and see other sizes. A somewhat more natural photo, with the tradeoff that it does show the limitations of the camera’s dynamic range. Oh and the layer mask method: whatta pain in the butt! I’m sure it’s possible to become efficient with practice, and it feels like a very powerful method viz the control you have over what comes out of it. Maybe with a truly exceptional shot that can’t be processed any other way it’d be worth the effort.