• Blogs
• Forums
• Tradeshows
• Partners
• White Papers
• Events

 

Shoot Expertise: That Film Look

Jun 27, 2006 8:01 AM, By Jan Ozer

Putting 24-frame camcorder footage to the test on image resolution and motion smoothness.

Sometimes we write to be definitive; sometimes we write simply to add to the body of knowledge growing about a particular subject. This article falls into the second category, relating to the ability of three camcorders — the Sony HDR-FX1, Canon's XL H1, and JVC's GY-HD100 — to capture video that looks like film.

Obviously, the “filmic” look has many characteristics, including resolution, frame rate, color, grain, lighting, depth of field, and many others. Here we focused on two of the more objective metrics: image resolution and motion smoothness.

Image resolution

Of the three camcorders, the JVC is the only one with a full progressive mode at 24fps, so it should be ideal for producing complete frames at 24fps. Both the Canon and the Sony use pseudo-progressive modes at 24fps to emulate the same result. Sony's technology is called CineFrame, and it's been blamed for everything from global warming to stuttering frames with poor resolution. Canon's recently introduced Frame mode uses a different technology and has received mixed reviews.

Critics of these pseudo-progressive approaches theorize that because the Sony and Canon camcorders have interlaced CCDs, they must merge the two fields to produce the progressive frame, reducing the resolution of the final frame. To test this, I used the resolution test pattern from DSC Labs' CamBook4 (www.dsclabs.com), shooting all three cameras in their respective 24fps progressive modes and comparing the results.

For the record, I captured all video files and created all screen shots in Adobe Premiere Pro using presets and the intermediate AVI codec in CineForm's Aspect HD product line. In testing resolution, I considered two scenarios, filmmakers who sought to convert their video to film and DVD and other producers seeking a film-like look for digital distribution.

For the first group, I compared images at full resolution. The problem here, of course, is that both the Canon and Sony capture video at a display resolution of 1920×1080, while the 720p JVC captures at 1280×720. Reasoning that when producing film, producers would have to scale the different formats to the same output resolution, I scaled the JVC frame to 1920×1080 resolution in Premiere and rendered the test images.

The results were mixed, as shown in Figure 2. In terms of vertical resolution, the Canon was the clear winner, showing the most definition around the 500 lines per inch block near the top of all charts. Sony's image showed slightly higher resolution than the JVC at 500 lines per inch, but was also visibly less crisp. Interestingly, in terms of horizontal resolution, which the two 1080i cameras should have won hands down, the JVC showed the best definition in the 500 lines per inch block on the bottom of all frames, though the results were very close.

To drill down into the HDR-FX1's performance, I compared frames shot in 24fps CineFrame mode with frames shot in normal 1080i mode, with the results shown in Figure 3. In vertical resolution, where I would expect to see a difference, the 60i frame is noticeably clearer than the CineFrame mode. Horizontal resolution is pretty much identical.

Interestingly, when we perform the same analysis on the XL H1, shown in Figure 4, the results are virtually identical. This suggests that Canon is using a different technique than Sony to produce progressive frames, with measurably better results.

That said, the results in real-world tests were much closer. For example, Figure 1 shows comparative frames from footage of a statue. This is a very challenging frame because of the detail in the hedges behind the statue; all three frames show blockiness upon close inspection. Quality differences here relate more to color balancing than any gross difference in resolution.

The resolution chart and statue shots were low-motion shots, and I wondered if higher-motion sequences would produce noticeable differences among the camcorders. Several of the clips shot for these tests involved more significant motion, including that shown in Figure 5, of singer Rene Marie warming up for her recent concert in Galax, Va.

Despite the motion, all edges in all three cameras are visibly sharp. None of the three camcorders appears to have a significant edge in image quality, at least not at full resolution.

How about the wedding or event videographers who don't want to render to film but want a filmlike look on DVD? In Figure 6, compression to typical DVD parameters appears to minimize any qualitative differences between the frames, and other post-compression comparisons were similar. All of the technologies seem relatively close in individual frame quality.

Motion Smoothness

The technical rap on the FX1 is that, when subsampling down from 60i video to 24p, it produced choppy video with irregular motion. To test this, I studied our non-resolution chart clips, which included warmup scenes from a jazz concert, a ballerina practicing jumps, a long slow zoom into a statue, and scenes of wind blowing through the branches of a crabapple tree. I played all videos side by side, both on computers and on a DVD player to a television set, comparing the results to the JVC video.

To my eyes, the Sony video showed some slight pulsing during the slow zoom sequence, but during other scenes, like the spinning ballerina or blowing tree branches, the motion did look perfectly natural. On the DVD, there was noise at the bottom and top of the video, where it looked like a line of video was shifting one pixel up and down for each complete frame, an artifact that wasn't present in video from the other two camcorders.

The video produced by the Canon was very similar to the JVC. Those interested in viewing this video can access low-resolution streaming files or purchase a DVD with these sample clips encoded into MPEG-2 format and details on the original captured video files at www.doceo.com.

Summary

Where does that leave us? Though the difference was less than I expected, the lower-resolution still images and artifacts noted during realtime playback convince us Sony's FX1 CineFrame is not the best option for produced-for-film or film-like looks on DVD. (However, producers have successfully shot natively and posted to film as well as achieved a film look for DVD via post.) If you own an FX1, however, don't expect a tremendous increase in quality if you buy or rent the JVC or Canon unit.

Canon's 24 Frame mode proved very competent in our trials, with crisp image quality and smooth playback. It's also easy to use and shoot and good for general 60i production as well, both advantages that the JVC doesn't share. However, its almost-$10,000 price tag is intimidating in this class of cameras. That leaves the JVC, which costs $3,000 less than the Canon XL H1 and offers true progressive frames, good resolution, and smoothness throughout.

---

To comment on this article, email the Digital Content Producer staff at dcpfeedback@prismb2b.com.

© 2008 Penton Media, Inc.