Canon HG10 Camcorder Review
The Canon HR10 is a looker, both in person and by the specs sheet. Despite the first class pedigree of the imaging system (identical to the HV20), we were still skeptical of the AVCHD codec, which has given other manufacturers problems with motion. Aside from some issues with 24P performance, it appears Canon has succeeded again. The HG10 tops the competition in performance, portability, and handling. Read on to see how.
Video Performance* (9.75)*
The Canon HG10 comes equipped with a 1/2.7' CMOS sensor, which has a gross pixel count of 2,960,000. The effective pixel count in Video mode is approximately 2,070,000. Working magic behind the imager is the Digic DVII processor, which is actually two processors: one for video and one for stills. This is the same lens/imager/processor system found on the Canon HV20, their top consumer HDV camcorder, and the HR10, their AVCHD camcorder that records to DVD. While we haven’t yet tested the HR10, the HV20 absolutely floored us with its performance, and will almost certainly walk away with our coveted Camcorder of the Year award (decided in October).
The HG10 has some big shoes to fill, given our love for the HV20, and some obstacles to overcome, given the poor showing AVCHD has shown us so far. We’re pleased to say that in many instances, it’s difficult to tell the difference between the HV20 and the HG10. First of all, the color performance looks absolutely identical, which is a huge score for Canon. Of all the HD camcorders we’ve tested this year, Canon has outperformed them all in color. Sharpness in static and very slow moving shots is where we first started to see the difference between the two. Though our resolution test had the HG10 scoring slightly higher than the HV20, to the eye the HG10 lacks some of the very fine detail. Only the sharp-eyed viewer is going to see the difference on your big screen HDTV. But then again, you wouldn’t have purchased the expensive camcorder and the HDTV unless you were a sharp-eyed viewer.
Every manufacturer tweaks their AVCHD compression a little differently. We have not been impressed to date with the ability of AVCHD to handle motion. In most instances, we’ve seen trailing or compression artifacts along the borders of moving subjects. Once again, we’re impressed, but not to the degree we were with color performance. There is still an issue with moving objects, but the trailing is not as harsh as it was on the Panasonic HDC-SD1 or the Sony HDR-SR1. This holds true only for 60i video. 24P was a different story.
Switching to 24P in bright light, the color performance and apparent resolution are more or less identical to the 1080/60i. Obviously, Canon is going to tout any consumer camcorder with true 24P. In the case of the HV20, they were absolutely right to do so – the video looked beautiful and managed to side-step the staccato look of older camcorders with a "24P-like effect."
Not so with the HG10. For whatever reason (and we have our theories), the 24P video on the HG10 looks terrible. Both panning and tilting create a very choppy movement, even when moving slowly. This is not the motion blur you might get when shooting at a slow shutter speed in 60i. It’s a stuttering, chug-chug-chug effect that absolutely kills the shot. Our best guess is that the progressive scan and output is in conflict with the AVCHD compression, which is extremely taxing on the processor. HDV is less processor-intensive, which may be why it looked so much better on the HV20. Regardless of the reason, we don’t recommend shooting in 24P unless you’re going for a special effect.
Cine mode is a gamma setting that affects color behavior in the Canon HG10. It can be used in 60i or 24P. The HG10 groups the Cine mode with the AE mode settings, which unfortunately prevents it from being used at the same time as Aperture or Shutter Priority mode.
This graph illustrates how the gamma curves differ in normal and Cine mode (Cine mode corresponds to the Cine 1 setting on the Canon XH A1). Cine mode is meant to emulate film tones. There is less range in the shadows and highlights, which has good and bad points. Hot spots are less likely to blow out, but shadows lose a lot of detail. The idea here is that if you’re going for a film look, you’d better take the time to light your scene for film. Properly done, the mid-tones absolutely sing. Saturation is reduced, and detail in the mid-tones pops. This setting is absolutely worth playing with if you’re going for a more professional look.
Low Light Performance* (8.47)*
The large 1/2.7" CMOS sensor should have a big benefit for low light performance. Some manufacturers, including Panasonic and JVC, opt for three smaller sensors in their high-end consumer camcorders, attempting to produce better color in bright light and hoping to slide by in low light. The results of our Great HD Shootout earlier this year indicate that all did not go as planned. In fact, the Canon HV20 (single chip) and the Panasonic HDC-SD1 (three chip) had at least equal color performance in bright light, with Canon easily topping the competition in low light. Both were better than the JVC GZ-HD7, which tended to warm the colors too much.
In 60i at 60 lux, the image produced some surprising results. Held up side-by-side with the Canon HV20, the HG10 actually looked less noisy, to a significant degree. When we tested the HV20, there was a fair amount of fine grain noise. The HG10 showed almost none. However, the apparent sharpness of the HG10 was reduced compared to the HV10, almost in direct proportion to the decrease in noise. Because the imager and processor are identical in these camcorders and these images were shot under identical conditions, we’re guessing that as the HG10 loses information in the AVCHD compression process and creates artifacts, it’s also masking some of the noise. When all is said and done, it’s not a terrible side effect. We wish all lossy compression was this beneficial.
Lowering the shutter speed down to 1/30th helped even more, though the blur here became quite noticeable.
15 lux in Auto mode was, as usual, the great leveler. At this low light level, the Canon HV20 and the HG10 showed equal amounts of noise, and the footage was not very pleasant to look at.
15 lux at a 1/30th shutter speed did not seem to have much of an impact on color performance or fine detail, except to reduce noise.
Switching to 24P brightens in low light brightens the image because the default shutter speed is slower (1/48th versus 1/60th). If you read the Video Performance section above, though, you already know what we’re going to say about 24P on this camcorder. The video stutters so much during any movement that the footage is virtually unwatchable.
This is a huge blow to low light performance. The HV20’s 24P mode allowed it to reach some incredible sensitivity (producing 50 IRE at only 3 lux). With 24P mode out of the game, the HG10 is no longer in the running for superstar – at least not to the same degree as the HV20.
Additional testing for low light performance entails shooting Gretag MacBeth Color Checker chart at 60 lux, then running frame grabs through Imatest imaging software to determine color accuracy, noise, and saturation. The second part of the test involves continually lowering the light levels until the camcorder can produce a maximum IRE of 50. At best, the Canon HG10 was able to produce 50 IRE at 8 lux when shooting in 1080/60i. The HV20 was statistically the same at 7 lux. In the case of the HV20, we ultimately factored the 24P performance into the score, which was a much more impressive 3 lux. However, the 24P performance on the HG10 looked so bad that it would be unfair to use scores derived from that mode. At 60 lux, the HG10 produced a color error of 10.7, with a noise level of 0.9125 percent and a saturation of 82.39. These scores were slightly better than the HV20 and the Sony HDR-HC7.
Canon equips its camcorders with Super Range OIS, an optical image stabilization (OIS) system that decreases the effects of shaking the camera while recording video. The HG10’s OIS was tested using our custom-built shake emulator at two different speeds. The first speed simulates a light walk with minimal shaking. Speed two simulates a more unstable shot, similar to a bumpy car ride or jogging with the camera.
At speed one, we found the HG10’s OIS performs as well as its HDV equivalent, Canon’s HV20, with a 86 percent shake reduction. The HG10 did significantly better than the HV20 at speed two, with a 50 percent shake reduction. Interestingly enough, we discovered our simulation process allowed for a remarkable effect. After approximately 20 seconds on speed two, the Canon HG10’s OIS system could process the equal gyrating motion of our machine and produce a much better shake reduction of 80 percent. Despite the fact that the camcorder appears to have the same learning curve as a raptor, we had to discard the score. This result does not seem feasible when recording outside of a simulation room, as most shaking motions would not maintain the same amplitude and frequency over long periods of time.
Wide Angle* (9.6)
*We tested the HG10’s maximum field of view by pulling the zoom all the way back to the widest angle, turning off OIS, and placing it on a tripod in front of a wall. The right and left angles were then measured by a vertical laser, and the difference between the angle measurements gave us a maximum field of view. We also viewed the full video frame on an external monitor to make the calculations, as some LCD screens do not show the full recorded frame. From this we derived that the HG10’s maximum field of view is 48 degrees.