HD100 Gamma/Colour Tests + DVX100 Match Settings

[dashwood]

I just spent many hours thoroughly scoping both the DVX100A and the HD100.
You can look at the results in the following quicktime files (advance through frame-by-frame to compare results.)

I also have noted my "wide-latitude," "low-light," and "filmout" settings.

HD100 Gamma and Matrix settings:
http://homepage.mac.com/timdashwood/.Public/HD100-DVX_GAMMA-MATRIX/HD100_GAMMA-MATRIX-Sorenson.mov

DVX100 Gamma Settings:
http://homepage.mac.com/timdashwood/.Public/HD100-DVX_GAMMA-MATRIX/DVX100-GAMMA-Sorenson.mov

DVX100 Colour Matrix Settings:
http://homepage.mac.com/timdashwood/.Public/HD100-DVX_GAMMA-MATRIX/DVX100-MATRIX-Sorenson.mov


Here's my technical matched settings for the three typical DVX100 setups. I got them as close as possible. Please note that these comparisons are based on the DVX100A with a setup of 7.5IRE. The HD100 was also setup with 7.5, but I have confirmed it is only added on the analog output. The 1394 is 0 setup.


Downloadable pdf of 6 of my scene file recipes:
http://homepage.mac.com/timdashwood/.Public/HD100_Scene_File_Recipes1.pdf

JPG:
http://homepage.mac.com/timdashwood/.Public/HD100_Scene_File_Recipes1.jpg

The colour chart used is a standard GretagMacbeth ColorChecker. I needed a smooth grad to determine "stepping" in the Mpeg compression, so I created the grad card myself in photoshop using a linear grad from absolute black to absolute white. I printed it on a good HP injet photo printer.

Before I started these tests I reset the HD100 to factory defaults and initialized scene file 6 on the DVX. I set the HD100 to DV24PA mode as well, 4x3. The HD100 colour gain was turned off (B&W) and the DVX100A was turned down as much as possible to give me the thinnest line possible on the waveform monitor. The line is much thicker on the DVX100A tests because it can't totally remove saturation, so you are seeing a bit of RGB response.
The HD100 was set to 108% clip to extend the recording latitude, and it seems from the scope results that the DVX100 also records 108% super-white.

Next, I white balanced and set a 18% Gray Card (on angle to light source) to 50IRE for each camera - using the default out-of-box scene files.
This gave me ƒ2.4 for the DVX and ƒ1.4 for the HD100. I had both cameras simultaneously plugged into my analog waveform/vectorscope so I could confirm the digital results via firewire. I set both cameras to 7.5IRE because I'm in North America, but I observed that the DVX100A passed the 7.5 setup through the digital FW cable, but the HD100 does not. I took this into account when I eventually created the DVX match settings.

I then went through all of the gamma combinations I could think of on the DVX and then did all of the colour matrix combinations. On the HD100 I went through a huge amount of combinations of curve settings, but only compared the cine and standard matrix settings to each other.

To determine matched settings for the DVX100A, I brought up the frame grabs for Cinelike, Cinelike_D, and Cinelike_V and determined how to match the shape in the HD100. I then opened the vectorscope frame grabs and tweaked the colour settings to get them as close as possible. It wasn't easy, and my resultant "recipes" are not perfect, but they are close enough to do a multicamera shoot and then only require slight adjustments in post.

It is pretty amazing the depth of control available in the HD100... I didn't even come close to dialing in ALL of the possible combinations, but I think my tests demonstrate the effects of each control well enough to create some cool looks.

I think that I will typically be shooting "old school" style with my wide-latitude setting to capture as much info as possible in my "digital neg" and then leave the colour correction to post production. I confirmed (as reported) that the MPEG compression creates bigger "steps" in luminance in the blacks than in the whites. That is why I increased the "master black" levels in my wide-latitude setting to get the most out of the compression. In this case I would expose for my highlights and squeeze about 8 stops of latitude between 15IRE and 100IRE.

Mac users can open my .mac public folder (timdashwood) and drag the JPEG frame grabs to your own hard drive. The PC guys can either download the iDisk utility or go here (http://homepage.mac.com/timdashwood/public) and download them individually from the HD100-DVX_GAMMA-MATRIX folder.


Tim

[magji2000]

Great information thats for the info

[Isaac_Brody]

Wow Tim. Good work. :thumbsup:

[Barry_Green]

... and this is why Tim Dashwood is one of the "must-read" posters!

[WarrenS]

I can't wait to try 'em out. Thanks Tim!

For those of us who haven't used the DVX (me), would you educate us when you might use these settings?

[dashwood]

I can't wait to try 'em out. Thanks Tim!

For those of us who haven't used the DVX (me), would you educate us when you might use these settings?


Reason #1: So many people on this forum have complained that the HD100 cine matrix doesn't look as "vibrant" out-of-the-box as the DVX100 cine matrix everyone is accustomed to. I decided we might as well level the playing field by matching the look so many people seem to like. Personally, I prefer to shoot with the slightly muted HD100 ciné matrix and colour correct in post. I think it is a safer option considering the MPEG encoder and 4:2:0 colour sampling.

Reason #2: If you plan to do a multicamera shoot DV with the HD100 and a DVX100A as a "B" or "C" camera, then you can match them all and make it a little bit easier on your online editor/colourist.

Reason #3: This still is DVXUSER.com, so we should try to maintain the DVX as the lowest common denominator when comparing any other cameras. This helps people put things like gamma and colour response in perspective to what they are used to.

Tim

[WarrenS]

Thanks for the info, Tim.

[Spiff_2]

Just a quick comment:


I set the HD100 to DV24PA mode as well, 4x3.

If you set the JVC HD100 to the DV mode for all your colour matrix settings, you may find when shooting HD that the values provided in this test don't work, on account of the the different colour spaces recorded by HDV vs. DV. That is to say YUV.7xx (HDV) vs. YUV.6xx (DV).

Perhaps dashwood can confirm/refute this statement?

-Spiff

[dashwood]

Just a quick comment:



If you set the JVC HD100 to the DV mode for all your colour matrix settings, you may find when shooting HD that the values provided in this test don't work, on account of the the different colour spaces recorded by HDV vs. DV. That is to say YUV.7xx (HDV) vs. YUV.6xx (DV).

Perhaps dashwood can confirm/refute this statement?

-Spiff

That is true, but in the case of a DVX100A match, the idea would be for them both to shoot DV.

In the case of the HD100 curves themselves, the controls will still create the same luma response curves in relation to each other. There may be some differences in certain hues (reds) when in HDV mode.

Interestingly, in my firmware version it seems that the MPEG2 encoder is upstream of the DV encoder even in DV mode. I've been told (2nd hand info) that sometimes that happens in the older firmware and the MPEG encoder doesn't turn off, but has been corrected in the latest firmware.
I first observed this phenomenon on day1 (http://www.dvxuser.com/V3/showpost.php?p=275038&postcount=33) when I did a quick test in DV mode and saw ghost artifacting, trails, & bad MPEG artifacts on motion, etc.

Tim

[KCFilms]

It will be interesting to see how footage from the HD100 holds up in color correction.

I seem to recall Nancy Schreiber always trying to get as close as possible to the final look she intended, in the field (while shooting "November.") She, aparently, could see a degridation in the footage that had been color corrected over shots that were not.

In my experience, I've noticed DV artifacts become more noticeable after color correction (especially in Final Cut Pro, less so in A.E.)

[stephenlnoe]

It will be interesting to see how footage from the HD100 holds up in color correction....


I've been working extensively with color correction on the HD-100 clips. They hold up very, very well. I've done full color composites, title composites, skin tone adjustments, selective color hue adjustments, selective color desaturations, pulled blacks up and down, histomatched against all kinds of footage. JVC's pulled off a great iteration of HDV.

I am a big advocate of getting it right in camera, however, It's nice to know you can pull down the blacks (in post) with the JVC footage and not have the image fall apart. Uprezzing to 1080p has been terrific as well. This aspect of the HD-100's capabilities is first rate. The blacks (under gain) could be a bit less noisy but all in all it is terrific.

[Robert_Niemann]

The HD100 has the capability of uprezzing to 1080p? And what is the profit of 1080p by using a camera, that can only reord 720p?

[stephenlnoe]

The HD100 has the capability of uprezzing to 1080p? And what is the profit of 1080p by using a camera, that can only reord 720p?
My direct purpose of uprezzing to 1080p was to see how well the image stretched and remained clear and useable at a much higher scale. This simulates projection...

[Barry_Green]

The HD100 has the capability of uprezzing to 1080p?

No, the camera can't. Although it can cross-convert its signal to 1080i.

[mezelf27]

The HD100 has the capability of uprezzing to 1080p? And what is the profit of 1080p by using a camera, that can only reord 720p?

There could be loads of reasons for that. Remember that the pixel-count of the upcoming HVX200 (probably) won't be any better than 720p, but it outputs 1080p to (to card as well as output)...

[Mediacre]

I'm not a DVX100a user. Could anybody tell me what's the difference between cinelike, cinelike d and cinelike v?

[Barry_Green]

CineLike is the original. CineLike_D is designed to stretch the latitude for the most dynamic range (_D = dynamic). CineLike_V is for contrastier, punchier colors for video screens (_V = video).

[dashwood]

CineLike is the original. CineLike_D is designed to stretch the latitude for the most dynamic range (_D = dynamic). CineLike_V is for contrastier, punchier colors for video screens (_V = video).

Barry, did Panasonic change the main "Cinelike" curve between the DVX100 and DVX100A? My understanding of the Cinelike curve in the original DVX100 was that it was considered a "filmout" 1:1 curve. It seemed to clip the whites very early, but the Cinelike curve in the DVX100A doesn't seem to be so bad.

[Barry_Green]

As I understand it CineLike is unchanged. They kept the DVX100 gammas the same so that footage could better intercut.

[Chance White]

Wow... awesome to the max! Thank you.