Photography links (back)

5d2 DCF?

Here's my workflow up to now:
Raw data are categorized in a "assets" folder with sub-directories according to date and location, with all associated data (pictures, video, gpx, sound files etc) in the subdirectory.

For images, I capture both jpg at high resolution and cr2 (although this may change soon, since I really don't see the advantage of having jpg for what I do - but what a waste of hardware). Up to photo nr. 9999 images were named img_xxxx, but since the numbering scheme rolled back after that I now name images img_year_month_day_xxxx which I believe will make it easier to backtrack as well. This rule is set up in Canon's EOS Utility, that I use to download the images (but any rename utility could be used).

Video is also downloaded the same way, but stored in a "mov" subfolder.

For gpx tagging, currently I'm using the "geologtag" app for the Iphone 3Gs. This works well and is fairly accurate (but no bells and whistles), and the files can be downloaded using a web browser. However, the app must be in the foreground to record (not sure how OS4 multitasking will change this). The app will also prevent the iphone going to sleep while it's in the foreground. I'm not deligent in recording gpx data.

For sound files, I use the iprorecorder app for the Iphone, which gives excellent results. This is mostly for video, and I've not used it much in production.

Image workflow

Once the raw image data is downloaded, the first step is to use RoboGeo to tag the image EXIF headers with the gpx data. They are then renamed using the freeware "ReNamer" to contain a "_gpx" flag in the filename. "ReNamer" can store "profiles" to keep everything consistent. Since not all images are necessarily tagged, I sort them by date in windows explorer and drop the recent ones in "ReNamer" and rename them using a stored profile. Both cr2 and jpg are processed.

Next step is to import into Adobe Lightroom. After this is done, some effort goes into stacking cr2 and jpg's together. "Auto-stack by capture date" (LR2) will accomplish this to some extend, but the smallest resolution is 1 second, which may stack photos taken with exposure bracketing together, or photos that were taken very soon after each other (like continuous exposure). I have no way around that. Also, this will stack with jpg's on top. Since I work with cr2's I like to have cr2 files on top, so - 1. expand all stacks, 2. select jpg only, 3. select all photos, 4. move down in stack. This will move cr2 files to the top. I also added three custom filters based on filename - jpg, cr2+tiff and "none". Once this is done, I can go through the filmstrip for a first round of rejection of the photos. Once I've marked all rejects as "rejected", I can select them (while all stacks are collapsed), and drag-and-drop them in LR2 to a "recycle" folder that I created on a disk that does not get backed up. This will move the rejected photos with all photos in the stack (the rest of the photos in the stack does not need to be marked rejected). Be sure to empty the "recycle" folder (by deleting the photos in LR2) so every now and then or Lightroom will crash.

The next step is to filter out unwanted photos from the stacks. this may be multiple exposures, brackets or what not. For this, I use the "cr2+tiff" filter and expand all stacks and work from there. I've not played much with HDR so I have no workflow for that. Photoshop get used for panoramas, doing a excellent job. Again all rejects are moved to the "recycle" folder.

For final processing, I use noise ninja, DXO optics pro (mostly for wide angle geometric correction up to 30mm) and Photoshop (mostly for panoramas). Any editing will extend the filename further with "_edit" or sometimes "_DXO" (althought dxo optics does not always add the extension, depending on if it's launched from Lightroom or on it's own). Files to the external editors gets exported as 16-bit per channel prophoto RGB, and get imported as 16 bit tiff with no layering. Obviously the filesizes get big, particularly for panoramas, and since they will be in the assets disk space, after they are exported as jpg files the tiff files gets deleted. One can always rebuild them if needed.

Once I'm happy with a image, I export jpeg files to a "media" folder on the same disk as the assets folder. Initially I export to a media/hold folder for further review. After that the image will be placed in a media subfolder where it's sorted according to location.

From the Media folder I generate the web pages seen here, using LR2, print using a Canon IP5200 with Canon's "easy print utility" and preview using Irfanview.

Work is done on a 3Ghz core2duo Intel based system with 8GB ram, using windows 7 pro 64 bit. Video card is a ATI 4950 (2GB).

Video workflow

1920x1080 Progressive video arrives from the 5d2 in .mov containers encoded in a very high quality implementation of the .h264 codec (at some 38MB/sec). To date it's been difficult to find windows-based non-linear editors that can handle this. But, it can be played back (at least on my system) without problems using the freeware VLC media player.

So, using VLC, I decide what I want to keep. The rest gets deleted. Since the 5d2 recently became 24p capable, If I use that format, I add a _24p tag after the filename. Otherwise it is assumed to 30p (or 29.95, which happens anyway after cineform is done). I do not mix 24p/30p, and I will probably not use 24p that much (except for 50Hz countries).

After that, I use NeoScene to transcode in "high" quality to a temporary "work" disk (no backup) using the minimal-loss cineform codec. Files now approximately double in size, but they become much easier to edit. In addition, the color space gets expanded. From there: Sony Vegas Pro 9 (64bit) for editing and VirtualDub (free) with the "AntiShake" plugin (free) to remove jitters. I still have to find a good solution to remove dust, altough I used the "wire removal" tool in aftereffects demo with good results. I could not get the Virtualdub "spotremover" plugin to work.

It is important to change the properties of the video imported into Sony Vegas Pro to "progressive", since by default it is assumed to be interlaced and if left uncorrected can cause soft edges. Once the first file is marked, all other files will be imported correctly and there's no need to do them one by one.

Once the video is done, I save it using the cineform codec, and then convert back to mp4 using ATI's Video converter, which does a excellent job, as long as you have a ATI video card. Otherwise the freeware "Super C" does a equally good, but much slower job. This then goes to Vimeo and get archived in the original "assets" disk under the "media" folder.

That's it!!!