Originally Posted by mrtickleuk
Which of the 4 different things in my post are you asking about? Many thanks to
for posting to give the name of the professional very expensive meter that I had in mind, when I made my general point
about how difficult it is to get bit-accurate YCrCb from devices, with errors and inaccuracies going unfixed for years and years.
It was a general
comment and in no way specific to the particular hardware of anything.
I think there is a distinction to be had between difficult and expensive, which is particularly poignant given the title and aims of this thread.
Making a 100% bit perfect pattern generator should be the easiest thing you can do. "All" you need is something that can drive serial IO fast enough that you can populate that IO with the correct pixel data and metadata. It's not hard to do. It is expensive, as general purpose devices (FPGAs) that could drive those speeds of serial IO are prohibitively expensive. An intermediate step is a device that can drive correctly formatted parallel data into an HDMI serializer (transmitter) chip, which is less expensive to do. Still expensive though (see Lumagen Radiance Pro, etc)
Lower cost devices are constrained by technical implementations where you may not have access to all the bits in the way you want, so compromises may have to be made. Those compromises might involve dithering, or not supporting some format, unavoidable internal conversions due to graphics architecture / pipeline limitations, etc.
The Murideo 6G product appears to be one such compromised device from what I can tell. I've seen photos of the inside. It doesn't have a 6G (600MHz) pattern generator in it from what I can tell, and this is the reason it can't support 4k DV at 60Hz, which a true 600MHz generator should be able to do. It has a Spartan 6 FPGA generating patterns, outputting parallel data to a 300MHz SIL9136 HDMI transmitter. It looks like that chip isn't actually connected to the HDMI connector if this box is capable of generating 600MHz timing, and is connected instead to a separate device that upscales the video output to 4K timing and is then connected to the connector (you can see the TMDS pairs leaving the anonymous device under a heatsink and going to the connector via the EMC filter components). So I guess at 4K60 10 bit what is actually being output by the generator chip is 1080p60 which is then scaled externally to 4K to overcome the limit in bandwidth (and hence the Dolby Vision limitations).
I guess the Murideo 6G is likely a victim of a company needing / wanting to get out a "6G" pattern generator fast and for little engineering effort, and is really a 300MHz generator bolted to something to make it look more-or-less 6G. The difficulty in making it accurate will come from the convolution in the output architecture with multiple stages that need to be made transparent.