[Michael D]

Hello Gene,

I see that you are back from vacation. I was wondering if you reviewed my two posts on 11 and 12 June in response to your last one on 10 June. I was hoping you would have further input and interest in this anomaly.

Michael.

[Michael D]

I too think there may be a bug in the Blackmagic codec, at least the codec embedded in their drivers (but apparently not with their standalone codec). Rather than repeat myself I would refer you to my posts (and responses) starting 8 June in the thread entitled

[Michael D]

I am also evaluating AJA mini-converters so let me tell you what I see as the differences between the D10A (component in only)and the D10AD (universal in). According to AJA

[Michael D]

Hello Gene,

For completeness, I am including the mean values of luma, red, green, and blue for a black and white file I captured using the DeckLink Pro card. I will limit my observations to a clip which I have labeled home.mov. As I previously stated, Blackmagic initially suggested I apply a black and white filter to the black and white files that exhibited the green tint. I applied the black and white file to home.mov within Premiere Pro and labeled the resulting clip homeprime.mov.

The mean values of the histograms of luma, red, green, and blue of a frame from home.mov with the DeckLink Pro card and drivers installed were 171, 172, 173, and 162, respectively. The red and green channels exhibited spikes and missing data. The gain of the blue channel was low and the clip had a green tint when viewed on the computer monitor and when viewed on the JVC monitor using the component out of the card. The green tint was much reduced when the clip was viewed on the JVC monitor using the composite out of the card.

The mean values of the histograms of luma, red, green, and blue of the same frame from homeprime.mov with the card and drivers installed were 170, 171, 170, and 170, respectively. All the channels now exhibited spikes and missing data. The clip appeared black and white when viewed on the computer monitor. The clip may have exhibited a very slight, almost undetectable green tint when viewed on the JVC monitor through the component out of the card (this may be the tint that is potentially inherent in CRT monitors that you and Bob Zelin mentioned). The clip appeared black and white when viewed on the JVC monitor through the composite of the card. Note that I could accomplish the same result as the black and white filter within Premiere Pro simply by increasing the gain of the blue channel within the After Effects plugin version of Color Finesse.

Now for some data with the drivers removed, the DeckLink Pro card physically removed, and the standalone Blackmagic codecs installed. The mean values of the histograms for luma, red, green, and blue for the clip home.mov were 164, 163, 165, and 162, respectively. The histograms did not exhibit spikes or missing data and the clip appeared black and white on the computer monitor.

The mean values of the histograms for luma, red, green, and blue for the clip homeprime.mov were 163, 162, 162, and 170, respectively. All of the channels exhibited spikes and missing data. The gain of the blue gain was greater than the gain of the red and green channels and the clip now appeared to have a blue tint on the computer monitor.

With the card and drivers in the system, the black and white files had to be color corrected by applying a black and white filter (think about the irony of that) which basically increased the gain of the blue channel. However, with the card and drivers removed, the gain of the blue channel was greater than the gain of the red and green channels and the clip now had a blue tint.

The tint I see on the black and white files is a function of potential inaccuracies in the gain of the red, green, and blue channels. The green tint I see has nothing to do with the JVC monitor (or any inherent almost undetectable tint due to the nature of CRTs).

Let me know your thoughts after you have a chance to review all this data. I feel that Blackmagic support has to step up to the plate and investigate whether there is any merit to the results of my analysis. I am aware of the problem and have developed a viable work around. You can

[Michael D]

Hello Gene.

I always thought that a lack of numerical precision was a potential cause of what I am experiencing. The spikes and missing data that I described as occurring in the red and green channels look like the artifacts in the conversion between Y

[Michael D]

Hello Gene,

The last sentence of the fourth paragraph from the end of my last post should read

[Michael D]

Hello Gene,

I am impressed. In just a few posts you hit the nail on the head and figured out that my concern goes beyond the green tint on the component out of the DeckLink Pro card. If only Blackmagic support had such insight. Let me expand on my previous comments by summarizing some results from my analysis of one of the color files I captured using the DeckLink Pro card.

I captured eight hours of digibeta files which were telecine transfers of 16mm color films. I will limit my observations to the third hour of capture, a file I have labeled nhour03.mov, and I will limit my observations to the mean values of luma, red, green, and blue using the software waveform monitor of the Color Finesse plugin within After Effects (and eliminate any influence of the JVC monitor).

I created a clip approximately 40 s long from the original capture nhour03.mov. I exported the clip to movie with the card and drivers in the system and labeled the clip card.mov. I also exported the clip to movie with the drivers removed, the card physically removed, and the standalone codecs installed and I labeled the clip code.mov.

Now for some numbers with the card and drivers installed. The mean values of luma, red, green, and blue for a frame from the original capture nhour03.mov were 116, 104, 120, and 127, respectively. The mean values for the same frame in the clip card.mov were 115, 104, 119, and 126, respectively (close enough for government work). The mean values for the same frame in the clip code.mov were also 115, 104, 119, and 126, respectively.

Now for some numbers with the drivers removed, the card physically removed, and the standalone codecs installed. I neglected to obtain the values for the original capture nhour03.mov. The mean values for luma, red, green, and blue for the frame from the clip card.mov were 107, 94, 110, and 126, respectively. The mean values for the frame from the clip code.mov were also 107, 94, 110, and 126, respectively.

The mean values of luma, red, and green for the clips card.mov and code.mov with the card and drivers installed are different than the mean values with the card and drivers removed and the standalone codecs installed. However, the mean value of the blue channel in both clips with the card and drivers installed is the same as the mean value in both clips with the card and drivers removed and the standalone codecs installed.

There

[Michael D]

Gene

Thank you for your continued interest. As I described in my last post (again with respect to my black and white files), the gain of the blue channel was less than the gain of the red and green channels when the DeckLink Pro card and drivers were installed but the gain of the blue channel was equal to the gain of the red and green channels when the card and drivers were removed. I had removed the JVC monitor from the system. What I didn

[Michael D]

Thank you for your reasoned response. I don

[Michael D]

Thank you for your interest. I don