Get 9.5 Stops of dynamic range from your DVX100

Pertierr
I can't see any DR difference inyour pics. I would love to see a good picture where the two pics are displayed on a split screen in order to avoid different display settings, and where we adjust the Andremedia channel for the same black level(by correction the offset) and noise level (by correcting the channel gain setting) as the DV output black level. Then we could evaluate the highlights in both channels. The conclusion wil be that there is no difference. The only difference will be that the highest level is being reached 2 to 2.5 steps earlier, but the value will be the same as the highest Andremedio value. The ratio between the lowest and highest levels determines the DR of a system. In order to evaluate what midrange expansion and highlight compression does as compared to the straightforward logarithmic conversion it would be great to see pics of real scenes in split screen mode.
 
André,

I'm not pro or cons Andromeda.

Just made quickly this picture. (Just stretched to get a split)

As you see in the black levels, Andromeda performs as well as DV.
And Andromeda performs better in Highlight levels.

comparatif7mv.png


ps : je ne vois pas où tu veux en venir, André ?
 
Andre:

I think you are confusing the dynamic range of the output format with the dynamic range of the aquisition itself. The actual absolute value of the pixels is irrelevant. What matters is what level of light intensity those values represent, and whether or not there are diferent values for a certain range of light levels.

In any case, if you do not think an image such as the one posted by hvpz represents increased dynamic range, then I fail to see what kind of image would.

Cheers,
Juan
 
I don't think I am confusing DR definitions. From the posted picture, where the output gains were not adjusted like I suggested. I see that the displayed standard DV range of light intensities is much higher than the Andromedia signal. The blacks look about equal, the whites are significantly brighter, so the overall channel DR ratio is strongly in favor of the straight DV channel (because of these non matching gain settings). If getting all the gray levels from your testpattern is your criterium for DR, then just lower the exposure, and you will get the full range of your test pattern's gray levels. Highlight compression and DR are different things in video.
 
Incorrect again.

If you lower exposure and reshoot the frame via both DV and Andromeda at the same time, Andromeda still shows more dynamic range, and shows bars that the DV footage doesn't see.

Again, this has nothing to do with gain, or color correcting images to make them look a specific way. This is a dynamic range test, and the image above shows 1.)The data originally digitized by the A/D converters and B.)The exact same data after it has been processed by the camera into DV. The A/D data transfers more dynamic range from the scene to the output than the DV does.

Andre, let's just agree to disagree. But there is no doubt in my mind that you are confused about what is going on here. I just don't know how else to explain it.

Cheers,
Juan
 
The andromeda can clearly discern more shades of brightness in the original source shot.

I think Andre's pointing towards the ultimate brightness vs. the ultimate darkness of the recorded frame, and saying that the ultimate range is the same (the contrast ratio between brightest and darkest). But what Juan is pointing out is that the Andromeda retains detail over more stops of exposure, where the DVX processor clips those top 2.5 stops to white. The Andromeda is capable of resolving differences across a broader exposure range.
 
The number of shades is indeed lower in the DVX (soft) white clipping but is still far more than the testpattern represents. Reducing the number of shades (256/color in an overall linear 8 bit system) by (soft)white clipping only reduces grayscale resolution (=number of shades) in the non clipped part to a lower amount, but doesn't affect dynamic range. There is such a thing as windowing in digital techniques. Again, the testpattern can be shown exactly (all stops) as show through the Andromedia channel, if the correct exposure is used. The risk for getting digital conturing with the DVX processing is somewhat higher because of the reduced grayscle resolution, but this doesn't mean that there is any reduction in the number of stops (not shades) which can be reproduced
 
Again, the testpattern can be shown exactly (all stops) as show through the Andromedia channel, if the correct exposure is used
Click to expand...
Well, see, that's the problem. No they can't. A generic grayscale could, but not THIS grayscale, because there's too much of a difference in brightness from the brightest chip to the darkest. If you brought down the exposure to compensate for the white clip, you'll find the dark chips crushing to black.

Juan originally sent me a file using a standard HD grayscale chart, and both the DVX and the Andromeda could easily resolve all the chips. So as a latitude test it was useless. I suggested that he set up the test with TWO grayscale charts in the same frame, one lit brightly, one in shadow, and that would give us many more shades of gray, which would clearly exceed the limits of the camera to resolve. Once we exceeded those limits, we could then determine whether one held more detail longer than the other (or, if you suggest, and I used to think, it was merely a matter of exposure).

Instead Juan devised this custom chart that clearly exceeds the capability of the DVX to resolve. Yet the Andromeda still does -- it differentiates detail between the brightest chips. It can hold detail in the highlights that the DVX clips to white. It can resolve an image across a greater spread of light to dark than the unmodified DVX can.
 
You're missing the point entirely.

The CCDs are capturing the exact same image. It is the same camera that is doing the capturing. He's capturing the DV image via the firewire port, and the Andromeda image through the Andromeda, but we're talking about the exact same source data. Both the DVX-processed firewire output, and the Andromeda frame, are coming from the exact same CCD data! The fact that there's two more strips of white that are easily discernible is proof positive that the Andromeda is RETAINING more image detail.

The internal processing of the DVX is pushing the whites to clip much sooner than they otherwise may need to do. It's obvious that the data is being resolved, because the Andromeda is showing what's actually on the CCDs. Andromeda just retains the information that the DVX is clipping.

So the Andromeda is capable of capturing extended-latitude scenes that the DVX otherwise clips/discards.

It is the same image. Capturing at a different exposure level would change both, equally -- but it still wouldn't change the fact that the andromeda is retaining two and a half stops of highlight detail that the DVX processing clips to white.
 
Barry plse explain what I am missing... Although english is not my native language I think I understand the test setup and the fact that both pics are from the same CCD at the exact same instant. My point is that the dynamic range is not different in both cases. I am not interested to go any further in this discussion which belongs to the general knowledge in the digital image processing field. I hope that somebody will do a test with optimized settings for the DVX to proof that exactly the same number of (unclipped) grayshades can be captired and yr last statement is completely wrong.
 
I would like to see some footage shot with the panasonic 16:9 anamorphic adapter.
To my knowledge none have been produced. It would be nice to see some.

kind regards.
 
andre:
I think Barry explained it best, but like I said before let's agree to disagree. I still think you are extremely confused.

dmbdvx:
Those are coming very soon. We're going to shoot some footage with the release version this week, i'll slap on the 16:9 adapter on there. I don't think it'll be worth it, however, seeing all the limitations it has. I think a lot of people get better results just by letterboxing footage, but I might be wrong.

Cheers,
Juan
 
pertierr said:
Those are coming very soon. We're going to shoot some footage with the release version this week, i'll slap on the 16:9 adapter on there. I don't think it'll be worth it, however, seeing all the limitations it has. I think a lot of people get better results just by letterboxing footage, but I might be wrong.
Click to expand...

I'm no where near as techno savy as the rest of you, but I think Juan is right. Besides, why would you want to put the adapter on there anyway? The image you get is 1540 X 984 which is well above the resolution needed to get great 16:9 SD footage. Maybe I'm missing something here ...
 
pertierr said:
andre:
I think Barry explained it best, but like I said before let's agree to disagree. I still think you are extremely confused.

dmbdvx:
Those are coming very soon. We're going to shoot some footage with the release version this week, i'll slap on the 16:9 adapter on there. I don't think it'll be worth it, however, seeing all the limitations it has. I think a lot of people get better results just by letterboxing footage, but I might be wrong.

Cheers,
Juan
Click to expand...

Thanks for the 16:9 footage with the adapter, I would really like to see it, as I shoot with the adapter often.

I know that shooting with the adapter on a dvx makes a differnece, sooo.....
with the adapter I do not get the black bands..

I understand about croping, but wanted to see what the adapter offers.?

Just curious

thanks
 
dmbdvx said:
with the adapter I do not get the black bands..
Click to expand...

You should not get the black bands without the adapter either. The only time, from my understanding, of when you will get the black bands is if you use the letterbox mode (Which records black bars ontop of the 4:3 footage) or if you are viewing 16:9 footage on a 4:3 TV. Using the adapter, or the squeeze mode, and creating a 16:9 DVD for playback on a 16:9 screen should yield no black bars, just a full screen image at 16:9. (You're image will be better with the adapter as you are not streching the image like you are in squeeze mode.)

BTW- I'm interested in see what the adapter adds to the output as well. The only thing I can think of it adding is a wider field of view along the horizontal axis ...

Thanks for testing the for us Juan.
 
Sorry to chime in late. I haven't read the full thread due to time. But which gamma setting was used on the DVX for this latitude test?
 
No problem at all. If there is anything else in particular you'd like to see let me know and we'll shoot it if we can.

What i'm worried about is just that the quality of the glass and overall design of the anamorphic from what i've seen isn't very good, and I think it might reduce the actual resolving power of the system. I think this indeed happens when you record to DV, but it makes a much smaller difference since the frame size is so small.

Cheers,
Juan
 
pertierr said:
I'd say it really all depends on how successful the concept is and what the users ask for. If the users consistently ask for a PC version then it will be done. :)
Click to expand...

Andromeda rocks! I had a similar idea but here you go they already made it.

One more here for a PC version. No, you don't have to do testing of all PC USB ports out there, just find a few that can handle it whether it's a PCI card for a desktop or a PCMCIA adapter for a laptop. Why does it have to be USB anyway? Why couldn't it be Firewire or Firewire 800. Here is another good one - SATA. In any case there's only a few USB chipsets even though brand names could be different. It would be a lot cheaper for people to buy the right adapter rather than switch platforms.

Thanks
Alex
 
You must log in or register to reply here.
Back
Top