Go Back   Steve's Digicams Forums > Digital Cameras (Point and Shoot) > Hybrid Still/Movie/MP3 Digicams

Reply
 
Thread Tools Search this Thread
Old Oct 7, 2008, 10:48 AM   #11
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Thanks. I would like to see the res charts.

Subjective landscape and fluffy bunny clips are still very good on these cameras, as low bit rate h264 cameras can handle test charts much more easily, but break down easily in real world complex scenes. Fur is one thing that causes macro blocking artifacts. I go for the most complex areas, and particularly the classical scratch and rub of the pets fur, that blocks up inter-frame codecs. I look for the little mistakes, because they turn out to be much more obvious on bigger screens (like the cheaper 200 inch projection displays, and others, we are moving towards eventually), or when you blink or see with peripheral vision (can detect high speed phone).

I can tell you that you will probably see block problems (blurry blobs in h264) in every 720p mode if facing a complex moving scene, or one with a wavy rippling body of water in it. The bit rates are not sufficient enough for the codec. They really need to go for 10mb+ for the 720p mode.

Most video sites are useless, about as much as you can tell is how bad the latitude in a camera is not. But download sites, and video sites like vimeo, allow you to post a native format clip as a downloadable link below the main link. I use these clips extensively for comparisons.


Thanks.
Wayne12 is offline   Reply With Quote
Old Oct 7, 2008, 11:39 AM   #12
Junior Member
 
Join Date: Oct 2008
Posts: 13
Default

Wayne12 wrote:
Quote:
Subjective landscape and fluffy bunny clips are still very good on these cameras, as low bit rate h264 cameras can handle test charts much more easily, but break down easily in real world complex scenes......snip......
Thanks for the perspective on this. I'll try to include some of those as well. I'm relatively new to that aspect of these digital output cameras. My background is mainly in Lasers, electro-optics, electronic and software design.

My main beef with these cameras and the resulting end product is the audio. Our main application is for video podcasting where the lighting, scene and motion are relatively static. We are currently working on producing a series of tutorials on surface mount assembly and soldering.

We've also been asked by a number of customers to recommend low end cameras for producing vidcasts that have decent audio quality, but don't want to have to do post-production to synch up separate video and audio. To date the only low cost one we've recommended much is something like the Canon Z930.

I would be interested in suggestions for methods to simulate some of these types of scenes that generate artifacts in a consistent and reproduceable manner. We design and prototype a wide range of electronic, electro-optical and electro-mechanical devices for a living.

We could do some type of variable speed linear and/or rotational motion platform and have different patterns/textures to try.

We could take this discussion off into a new topic to discuss how to really hammer on these cameras and codecs in a controlled manner and see how they respond.


Quote:
Most video sites are useless, about as much as you can tell is how bad the latitude in a camera is not. But download sites, and video sites like vimeo, allow you to post a native format clip as a downloadable link below the main link. I use these clips extensively for comparisons
.
I've registered with Vimeo and will post clips there.

Quote:
The HDHS is said to be 1.6mp. This number rang bells, and now i realise it was because there are a few cameras that I do not enjoy the performance of.
What sensor do you think this may have? If (read when) I succumb and decide to open it up will perhaps I can confirm this.
Chipguy is offline   Reply With Quote
Old Oct 7, 2008, 11:59 AM   #13
Moderator
 
fishycomics's Avatar
 
Join Date: Jul 2005
Location: NYC NY
Posts: 9,623
Default

LOL yep I just seen the old page and now I see the new page it must just have been done LOL




fishycomics is offline   Reply With Quote
Old Oct 7, 2008, 2:35 PM   #14
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

It is 4:26am, so my answer is a bit rough.

Chipguy wrote:
Quote:
Wayne12 wrote:
Quote:
Subjective landscape and fluffy bunny clips are still very good on these cameras, as low bit rate h264 cameras can handle test charts much more easily, but break down easily in real world complex scenes......snip......
Thanks for the perspective on this. I'll try to include some of those as well. I'm relatively new to that aspect of these digital output cameras. My background is mainly in Lasers, electro-optics, electronic and software design.
Ironic, I am actually planning on making a new laser display scanning design, but being a novice I could do with some perspective on this. I am gearing up for some virtual os development around a parallel embedded processor (intellasys.net). I used to be in the digital cinema camera development projects at dvinfo.net. I was interested in doing my own camera, but not being an electronic engineer and most people being unreliable there, I couldn't find a suitable partner. Even any camera out there it is possible to tap and record the uncompressed video stream (much better quality) only the couple of engineers there were worth talking with.

If you want to rip apart your camera, there is a couple,of threads here where they open up Aiptek and do stuff to the mic, and another where they add a custom standard lens mount. the lens portions on these things can have fine cabling.

"My main beef with these cameras and the resulting end product is the audio. Our main application is for video podcasting where the lighting, scene and motion are relatively static. We are currently working on producing a series of tutorials on surface mount assembly and soldering."

Rapid movement of the board or machinery might not be perfectly ideal, I would be interested how well ti handles it, they present a relatively complex scene. If this camerae does not work so well i can recommend better (canon has 24mb/s AVCHD cameras out, and Samsung has a cheaper HMX20c that uses a nice Sony sensor). See my comment on the sensor below.

"We've also been asked by a number of customers to recommend low end cameras for producing vidcasts that have decent audio quality, but don't want to have to do post-production to synch up separate video and audio. To date the only low cost one we've recommended much is something like the Canon Z930."

Auto focus might be a problem, so fix focus cameras might better. Aiptek, DXG 595V (remember the v, that is the better one) Sanyo HD1010, Sanyo 100 (unknown, have not done review of footage). Are the best around here. I don't know if the Sanyo HD1010 is any better than the DXG, but it has manual controls and a true full HD mode. The cheaper Samsung shares chip with sr11 and 12 have similar chip (if not the same) and better color and probably latitude, the Canon (can't remember the 24mb/s models but three of them, but the previouse models broadly on par with the Sonies). Casio has just released a new model with cmos sensor AVCHD and Mpeg2 variable (worth examining). Canon HV20 and 30 are tape based canons. These apparently leave the Sanyo for dead.

"I would be interested in suggestions for methods to simulate some of these types of scenes that generate artifacts in a consistent and reproducible manner.¬* We design and prototype a wide range of electronic, electro-optical and electro-mechanical devices for a living."

To tired, but we do everyday street scenes and low light with panning, it shows up stuff in the detail. Our concerns are different from yours, and I think you will be better served then we are with these cameras (as the static simple scenes renders better). One important thing is latitude, that will keep bright parts of the board from blowing out (but you can adjust lighting tom do that, but some sensors can suffer too much). In the bright outside scenes you can see the shadows go very dark/black, and the bright parts go white, even the Sky and sidewalks. A good camera has a the right blue sky, most every bright thing with color and detail, and details in the shows if not color and shading. In low light, a windowed room with light coming in from outside, and shoot of outside is another contrasty thing that can also show up noise in an inadequate setup. Street lighting at night, room light at night are other examples. A dull lightbulb (15w incandescent in a medium sized room also gives an idea). once you see enough examples from people you get a feel about how well it is doing in a scene.

Optical imperfections in the lens you would already know about. The Sanyo HD1000 (and some other camera I can't remember) had blue chromatic aberrations from very bright highlights in a certain direction (red in the other). That maybe annoying filming surface mount stuff.

"We could do some type of variable speed linear and/or rotational motion platform and have different patterns/textures to try."

he, He, that is what i suggested years ago at dvinfo. Bit late to be thinking about it,. so i can't give an analysis at the moment. Waves on water is a popular one to break older codecs, complex non predictive movements, like bushes in strong rings is another. Bit, the most basic problem around here, has been rapid panning. Apparently this is a challenge on many cameras, they stutter. i don't know how much a problem with the newer higher bit rate 1440*1080 cameras.


I remember what i was suggesting was something like spinning cross-hairs with a square lien around it. Just spin the resolution chart. If the codec can;'t predict it, it has to depend on the data rate (unless it uses that inefficiently). Examining hope easily ti breaks up and blobs (hides macro blocks).

But what is best to remember is the way these sorts of compression codecs break down. Non predicted Motion and complexity (like random noise) just suck up data-rate which if not there, causes drop in quality. In complexity is shaping (which i think they now use). these sorts of things can distort what you see on the circuit board. The basic key frame intra frame method (apart from shaping and resolutions dropping) is not narrow the difference between surrounding pixels, thus shifting color and contrast, reducing detail. This makes images less striking, more murky, and reduces details which might be undesirable. Another aspect is loss of resolution and fuzziness on details, and also other artifacts like image ringing. For your use it is probably best to give a test run, adjust lighting and test run again, you should get a personable idea.

"We could take this discussion off into a new topic to discuss how to really hammer on these cameras and codecs in a controlled manner and see how they respond."

Very busy on a document, so that is something i will have to avoid for now..

Quote:
Quote:
The HDHS is said to be 1.6mp. This number rang bells, and now i realise it was because there are a few cameras that I do not enjoy the performance of.
What sensor do you think this may have? If (read when) I succumb and decide to open it up will perhaps I can confirm this.
If you can get to the firmware section it might list it. I think it was the altasens altrachrome that i am thinking of. Altasens was a top rating professional sensor at it's introduction, and we got excited about the consumer version, but the cameras did not match up to our expectations. I used to look up sensors for the digital cinema camera, and 1.6mp was pretty unusual, I think Kodak could have some. usually they are 1.3mp, or 1 mpixel. But I could be wrong. There was some interpolation issues in the cameras I remember.
Wayne12 is offline   Reply With Quote
Old Oct 7, 2008, 8:01 PM   #15
Junior Member
 
Join Date: Oct 2008
Posts: 13
Default

I think the sensor in the HSHD is the Aptina (formerly Micron) MT9M002 CMOS sensor which is a part of their DevCam HD series of sensors. The specs and features align perfectly.

The flyer can be found at

http://www.aptina.com/assets/downloadDocument.do?id=168

1472 x 1096 active array
3.24mm x 2.41mm imaging area
Frame Rates 1280 x 720 60fps
1440 x 1080 30fps
Output LVDS serial or 12bit parallel
Max clock rate 49.5 MHz
Max data rate 99 Megapixels/sec
Dynamic range 70.1db
Responsivity 1.4 V/lux-sec (550 nm)

Aptina's reference design board uses an Ambarella A2 video processor

Aptina does not make it easy to download their full data sheets. I had to register, do email verification, log back in, agree to a long NDA then get told I still didn't have permission to view the document and please fill out a sales contact form.

I rarely use or buy anything from a company that paranoid about letting you see their product information. That kind of thing drives me nuts. I'm sure I'll have to come up with some creative potential use numbers and suffer convincingly through a sales call to be granted permission to view the holy document.




Chipguy is offline   Reply With Quote
Old Oct 8, 2008, 11:07 AM   #16
Senior Member
 
Join Date: Dec 2007
Posts: 1,153
Default

Chipguy,

I don't understand your test.

Are you using something like this? Here:

http://www.studio1productions.com/xlr-bp_pro.htm



How are you connecting your microphones to the camcorder's external microphone input jack?

It's difficult to know what you're really doing here.

I mean you write "condenser" and "dynamic" microphones are the types of microphones you've tested, but you haven't listed any specifics about these particular microphones and you haven't explained how you've connected those microphones.

Which microphone models are you using, for example?

A Shure SM57 is a dynamic microphone, but it features an XLR out and one would need a way to connect that to the input on the Aiptek camcorder.

I suppose I'm just hoping you could provide more specifics.
Private Idaho is offline   Reply With Quote
Old Oct 8, 2008, 3:47 PM   #17
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Chipguy wrote:
Quote:
I think the sensor in the HSHD is the Aptina (formerly Micron) MT9M002 CMOS sensor which is a part of their DevCam HD series of sensors. The specs and features align perfectly.

1472 x 1096 active array
3.24mm x 2.41mm imaging area
Frame Rates 1280 x 720¬*¬* 60fps
¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬*¬* 1440 x 1080 30fps
Output LVDS serial or 12bit parallel
Max clock rate¬* 49.5 MHz
Max data rate¬* 99 Megapixels/sec
Dynamic range 70.1db
Responsivity 1.4 V/lux-sec (550 nm)
Good find, they are quoting dynamic range now.
Wayne12 is offline   Reply With Quote
Old Oct 10, 2008, 1:10 PM   #18
Junior Member
 
Join Date: Oct 2008
Posts: 13
Default

I finally got a reply back from Aiptek support about the HSHD battery fit and discrepancies between the manual(s) and actual camera.

All I can say is it is seems like pulling teeth to get any useful information out of them.

Their first reply only addressed the battery fit problem and suggested I return it for a new one as this was not a "common" problem with this model. They did not address the manual discepancies or tell me how to check the firmware at all.

I replied that the battery problem was not a show stopper and had been fixed, but that as it is a new model they might want to know about the problem. (I didn't bring up the point that as a newer model, how they knew what was and wasn't a "common" problem). I asked if they could address the other issues I had inquired about and another discrepancy I noticed this week between my manual and the newer one posted online this week that lists QVGA as a video mode.

My reply this morning asserted the following

1. The HSHD does NOT have a Mic icon that comes on when an external mic is plugged in. The reason stated is that this is a "scaled down" unit with only basic features. It doesn't specify a scaled down version of what.

2. the Icon On/Off is in the Settings menu. I did finally find it

3. There is no histogram function - manual misprint (twice evidently)

4. No firmware updates available

5. QVGA is for the HD-1, not the HSHD - again, evidently a misprint

6. Absolutlely dead silent on the question about how to check firmware.

I guess its back to playing with button combinations.

Still no joy on trying to obtain information from Aptina about the MT9M002 sensor and nothing from Ambarella about the A2 chip. I haven't received any reply from Aptina after two sales requests. It sounds like in any event even if I did obtain any information I would be precluded by an NDA from sharing any of it.

Back to real work. Hopefully I'll have some time this weekend to make some audio measurements.
Chipguy is offline   Reply With Quote
Old Oct 10, 2008, 1:38 PM   #19
Senior Member
 
Join Date: Dec 2007
Posts: 1,153
Default

Could you reveal which models of mics you used to do your tests?
Private Idaho is offline   Reply With Quote
Old Oct 10, 2008, 1:38 PM   #20
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Chipguy, have you got any short clips of the 720p mode, I am eager to see them?

Thanks.
Wayne12 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 10:16 AM.