Go Back   Steve's Digicams Forums > Digital Cameras (Point and Shoot) > Sanyo

Reply
 
Thread Tools Search this Thread
Old Jun 19, 2008, 8:43 PM   #11
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Have reviewed some of the footage. We wont expect to see as much difference in modes with higher bit rate to resolution ratios Still, the bigger the screen the more recognisable the differences.

--

Room 1 - Fluorescent light dark scene

Is it my imagination, or does this scene look a lot better than I remember it. Quiet a few cameras have exhibited noise problems i=on this scene, did he upgrade his lighting setup, or reposition it? If not, the camera is doing much better than expected.

The noise level seems very good, you will notice on closeup the reflection shadow to the left, of the balls movement crawls a little. Is this the noise removal resting itself, or just natural reflection/shading?

The metal arms on the balls might show very minor rolling shutter . But there seems to be stepping on them as they go through their reflective arc, you can see glimpses, and more on the fuller arc when slowed down on closeup. Something minor to the average viewer (as long as it does not crop up all over picture and important objects. There is a sort of rough surface, is this caused by a natural rough surface on the arms, if so I have not really noticed before. Lastly, my favorite, the blur ring of death on the reflected lights on the shiny balls. This is what stopped me from buying the previouse HD1000, as the brighter the light coming through the lens the bigger and worse it seemed. But here it seems minor, and maybe it is caused by the reflective scattering of the materials used in the balls themselves (or the frequencies of the light source) both of which are not caused by the camera. Will have to wait to I get to review the outdoor footage (if I can get them to run).

I notice some burn out though on the closeup though. the highlights seems to be just a little too gleamy. There is something about the color too, I have trouble putting my dull sleepy mind on. It is only a minor difference, but if it was not there it would make the colors look more charming to a consumer. But still, this does a great job, and I think a firmware adjustment could solve both these minor issues.

I think good job.
Wayne12 is offline   Reply With Quote
Old Jun 19, 2008, 8:44 PM   #12
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

http://www.watch.impress.co.jp/av/do...0618/hdshq.mp4

This file looked dark outside, I mean dark, even for being overcast. Maybe the exposure was not aggressive enough but should have adjustable exposure (right guys)?

However, the water is not the usual variety of wavy water we get from this reviewer, but illustrates something. The water undulating surface looks strange, almost like what we see with the shifting in a aggressive de-noise filtering, but it should not be anything to do with that. if this is the sensor I think it is, then this might be the problem I have seen in footage from another camera that uses that sensor. But I will have to examine closer.

Are the green strange and yellowish, I am not familiar with these trees and grasses? Looking at the colors in Room 1, the yellow fluorescent pen seems to stick out a lot more. Can anybody comment if they think any of the other colors are muted (this could be caused by the way information is processed in low light/fluorescent but this scene has been used before). The green liquid does not seam to be as bright as I remember (he could have changed that too).

Are any of the other pens muted in color, the green liquid looks different to me, anybody can common to that.

Warning, this should not happen but the colors of the hdshq.mp4 file in VLC player seem a bit more lively than on the Media Player Classic. It seems to be opposite on the hdhr.mp4 file. Maybe I have two different H264 renders in my system or something causing difference between players, and codec choice for particular resolutions.

http://www.watch.impress.co.jp/av/do...80618/hdhr.mp4
Wayne12 is offline   Reply With Quote
Old Jun 19, 2008, 8:47 PM   #13
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Room 2 - candlelit. Interesting, seems to be testing different low light modes and techniques against each other, in three candle light. Noise is there, but not too bad for consumer. I prefer much lower noise in such a scene, and I think might too. The days of lower than 1 lux cameras seems to be gone for now? Not much to say, as I don't know what he did, but colors seem richer and the candles seem to be blown out. Seems to be some rolling shutter effect with long exposure.

http://www.watch.impress.co.jp/av/do...0618/room2.mov

Conclusion:

The blown out highlights and slight color adjustment are the only two main concerns for me. Have yet to verify if that blue chromatic flare is on bright highlights. I like balance of image, that you look at and say is nice, even cozy (nice latitude, color balance and low noise) nothing harsh in balance, but with higher contrast if adjusted by exposure/scene select.

Hopefully I can get around to reviewing the main clip, and closeups, shortly.
Wayne12 is offline   Reply With Quote
Old Jun 19, 2008, 10:56 PM   #14
Senior Member
 
Join Date: Oct 2006
Posts: 584
Default

Wayne, to emphasize the lack of quality control in the testing by this group, here is the video for the Canon HV20:
http://www.watch.impress.co.jp/av/do...0207/ezsm5.mov from the page:
http://209.85.171.104/translate_c?hl...7/zooma294.htm


It certainly looks like the HD1010 was put on "extra vivid," but then the HV20 may have been on "cine". In any case, you can see the 24p shutter speed in the ball motion, and the definite "black level" in the HD1010.

But I couldn't draw any definite conclusions.

Trevmar is offline   Reply With Quote
Old Jun 20, 2008, 12:58 AM   #15
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Color on the Canon definitely better, and it can handle lower light which has made the grey desk come out more, but lighting does seem different. The HD1010 color might possibly be going up and down with different modes, or the different exposures also.

Have located previouse hd1000 review for comparison:

http://209.85.171.104/translate_c?hl...006-17,GGGL:en

Previously he had converted the h264 files to HDV so people could view various files. I believe this introduced their own problems.

The color looks better in the in the HD1000 indoor balls shot, even though it maybe a bit over the top.
Wayne12 is offline   Reply With Quote
Old Jun 20, 2008, 3:47 AM   #16
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Have managed to compare the indoor swinging balls video from the HD1000 and the Hd1010, and looks like everything I mentioned on the HD1010 is improved over the HD1000, including noise, except the color looks less on the HD1010 (and seems to be less highlight burnout, but exposure is different). The color does not worry me if that is what it looks like in real life, and is not more brightly saturated.

However, I can focus better now, and there is some noise on the HD1010 indoor balls footage, to a significant degree, but still much better than the HD1000.

http://www.watch.impress.co.jp/av/do...926/ezroom.m2t


The sample footage file will take a lot longer to download and examine.
Wayne12 is offline   Reply With Quote
Old Jun 24, 2008, 3:25 AM   #17
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

There is blocking/fuzziness on simple scenes with little movement in them. It seems to need more data-rate. My VLC media info feature reports 1080 at 29...FPS for most of the clips, so must be 1080p30. I don't' know if the extra frame rate of 720p60 compresses better then the extra resolution and slower frame rate of the 1080p, it might be a little better, but even at 720p in the pond scene I still see artifacts in the water.

In the indoor balls scene, there is blocking on the balls in this simple scene. I can also see interpolation issues in the movement (movement also includes the bars and reflections). In one frame there is a lot of aliasing interpolation around a pen (I test many frames but still only small sample of all frames).

In the pond scene, I saw thin vertical branches light and dark horizontal strips, and also stair case look, not anti-aliasing stair-casing, maybe interpolation issue. Very dark looking trees.

Color aberration still exists, but not as noticeable, more acceptable for consumers.

Water surface still blocky (dirty pond in front of car park and trees scene). Was having trouble finding all of it until I realise that a big patch was actually blurred out by the encoder.

Going to the duck scene. The video looks over exposed and washed out. On ducks neck front, there are burned out feathers, and blocking near that. You also see this in other scenes like the cats nose. Blocking in water, and movement on webbed feet. There were some additional two notes on issues to do with this scene I can't read.

Pink and purple flowers burned out Even tip of stick to lower left of pink and white flowers, shaded just white.

First cloud scene too dark leading to trees being under exposed. This is not necessarily an error, but maybe related to the exposure level being selected to display the clouds clearly over the trees.

Patchy color in the sky. Blocks where clouds thin and transparent. Static clouds have blocks.

Sky through leaves aberration maybe gone, but need bright blue sky against dark leaves and wider aperture.

Is there some slight yellow ting going on in the greens?

speckle of backgrounds, can't remember, but maybe less than the HD1000.

Cat in grass being pampered. Too much blocking in movement. Some frames, strange finger print outlines and blocked (blurred around).

Last sky and tree shot looks better in thin branches interpolation/aliasing this time.

Have pictures from Aiptek HD for less than half the price. Like color and exposure a lot more. We could say a bit of a touch of Panavision color lifting, but pleasant. I would like to see a direct comparison shoot off between the two cameras in 720p60 mode (the strongest compression mode in the Aiptek, and second strongest in the Sanyo, the 720p30 being the strongest in the Sanyo).

I am disappointed, the first controlled shots, indoors looking out on a cloudy day, obviously were not enough to give an idea of the camera. It is better, and maybe if the exposure was toned down, and color adjusted it would give better results again (hopefully even for latitude) but the compression is still a let down, and the latitude. The burn out is strange, like it is clipped in some way, some hardware or software design, so maybe it can be changed.

If you want consumer quality then their are cameras on the street for less than $200 that give suitable quality (higher speed versions coming). But we have seen this on Japanese cameras in recent years, rather than buy in ambarella h264 codec technology the industry has gone with in house/Japanese solutions, and we have waited. Samsung also have a higher data rate 1080p/i mode only camcorder, likely using a 6mp+ Sony chip and ambarella codec for $899 (or was that less). Seems to have yellow tinge, maybe a bit much color, but I would not be surprised if it was superior in some ways. But there were comments about the Samsung camera test frames I had seen, that you could get a Canon HV20 on ebay for the same price, and that was preferable. So where does that leave other similarly priced cameras?

The Japanese camera market is caching up, but it has meant years of waiting for encoding performance to come up. The Chinese now are hear with Ambarella chips. Are they as good as the Sanyo, probably not, but close enough to wonder what to do. Despite everything, I have respect for Sanyo, and hope they can bring a Red Scarlet like consumer grade camera (the shape is already right) of the quality of the Sony Sr12, and newer models, with 18mb/s+ for all resolution modes with even better h264 compression. I would even volunteer to help them with this if they wanted to contact me.
Wayne12 is offline   Reply With Quote
Old Jun 24, 2008, 11:21 PM   #18
Senior Member
 
Join Date: Oct 2006
Posts: 584
Default

Wayne,
The Sanyo HD1000 had a superior joystick controller to the HV20. But the new HF100 has its joystick at the edge of the LCD, exactly where your second thumb is sitting, and erases that one benefit of the HD1000 User interface, IMO.

I sent back my HD1000 when I did some objective testing of its compression artifacts and blockiness. The Sanyo compressor was having difficulties by comparison with either the DivX5 of my Kodak V1073, or the AVCHD of my HF100. In particular, straight vertical edges of objects tended to tear. I could never decide if this was an EIS or compression thing, but in any case, it wasn't acceptable. The compression blockiness was noticable in my HD1000, and the performance when there was noise in an image was abysmal.

As somebody who bought their first Praktica SLR in 1971, I have seen photography evolve, and learned not to get sucked in by the "whiz bang" stuff. Ultimately it is the solid footage one generates which will survive.

For me, I now use the HF100 (my two HV20s currently sitting idle) with a Kodak V1073 in my pocket for those occasions when subjects might be intimidated by anything looking remotely like a camcorder...

Trevmar is offline   Reply With Quote
Old Jun 25, 2008, 1:21 AM   #19
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

I agree it is the footage, and the handling at that level of footage. Noise is a killer for higher compressed stuff, and movement too if the codec cannot handle it. Nose just robs data rate from the codec. You look at the Aiptek 720p cameras, and they basically eliminated noise, and then were able to shape the image.


I would however, like to ask you more about the objective amount of blockiness and other quality issues you saw between these cameras and any single frames you may still have. However, which modes did you use to compare on each. 1080i compresses worse per pixel than 1080p or 720p60 which compresses better than 720p30 (and higher shutter speed and more depth of field is better) and they all have different data rates that give different quality. I usually say 720p60/50 is a good all-round compromise (until they come out with high enough 1080p60/50 data rates) but I would like to know how all the progressive modes compare. I think that that Kodak model has nearly as much data rate for 720p30 as the Sanyo has for 1080p (hence my point about giving 720p60/50 and 720p30/25 on the Sanyo with 14mb/s data rate). They need to take off the chasity belt.

I am not too surprised that the Kodak Divx5 codec was passing it. Before Ambarella, many Asian hybrids had trouble with quality, some even putting mpeg2 data in Mpeg4 containers. I was hoping that Sanyo would give ambarella a run for it's money.

The sad fact is, with todays cheap storage market, that they probably can do very good 35-50mb/s Mpeg2 compression in a handheld like the Sanyo HD1010 with a circuit not much more complex, if not less. Even being able to offer lower bit rate H264 modes (14 mb/s+) and higher bit rate mpeg2 modes (35-50mb/s) on the same camera (JVC is starting to do this). Again, they need to take the chastity belt off to compete with the new competitors.
Wayne12 is offline   Reply With Quote
Old Jun 25, 2008, 5:57 AM   #20
Senior Member
 
Join Date: Oct 2006
Posts: 584
Default

Wayne,
Yesterday I bought two 16 Gig SDHC cards from Newegg for $49 each. They are the "eee PC series." They work great in cameras (of course). To me, data rate is unimportant, the higher the better. Storage is not a limit. The limit is imposed by the hardware/software on the cameras, not on my ability to handle it. I always pull-down and re-encode the video (often to HDV) prior to sending it to Vegas, as editors really don't like the B frames of H264. So the 14 Mbps of the Kodak codec is just as easy for me to handle as the 9 Mbps of HD-SHQ in the HD1000. Of course - the Kodak software limits its maximum recording time of 29 minutes - a limitation when compared to the 40 minutes maximum of the HD1000 HD-SHQ. However the HF100 records for nearly 3 hours at 24p 12 Mbps AVCHD without stopping (16Gig SDHC, extended BP819 battery).

The Kodak camera is a pain in the neck to use, but it is unobtrusive, can be used almost everywhere to obtain candid video snippets. The OIS is really nice The audio is surprisingly good. And it is (relatively) cheap.

All my important video photography is shot indoors. I am only interested in motion-blurred 30p or 24p, as the edited clips will end up on DVD or YouTube. When testing the codecs I used AVIsynth and Virtualdub to examine video artifacts frame by frame. Some of the toolchain I use is documented here:
http://www.hv20.com/showthread.php?t=11136

Although the first generation HD1000 performs well in decent lighting (ignoring codec issues), the lack of color and noise in the HD1000 at low lighting levels is just not in the same league as the HF100 or the V1073.

Using ND filters, I estimated that color rendition of the HD1000 was at least 4 f-stops worse than the HF100 and noise at least 2 f-stops worse. The Kodak's V1073 large 1/1.63 CCD sensor gives similar excellent low-light performance to the HF100 (Canon's tiny imager is amazing). Let's hope the HD1010 has improved performance in the low-light area. Sanyo have to get that right (IMO).


Trevmar is offline   Reply With Quote
 
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -5. The time now is 9:26 AM.