Greetings all,
This is going to be a bit of a technical deep dive to address a topic that has come up in several threads here on the forum recently. People who are familiar with Dolby Vision as implemented on Blu-ray Disc players may have heard the terms “FEL” and “MEL”, and these terms are now being used here when talking about Strato V. However, as I’ll explain below, these terms really don’t apply to Strato V, so I’d like to clear up any confusion.
First, four quick bits of background that are going to be important.
Because Profile 7 was designed for optical discs, and there was a desire to have the discs be backward-compatible with HDR10-only players, Profile 7 starts with an HDR10 “base layer” that can be played on any 4K player. The base layer is then augmented with an “enhancement layer” when played on a Dolby Vision-capable player. (Decoding that enhancement layer is the part that requires the second video decoding engine mentioned above.) The enhancement layer in Profile 7 contains two elements:
Some folks have said things like, “Strato V doesn’t have FEL,” or have likened Strato V’s Dolby Vision implementation to Profile 7 with MEL. But this statement is not correct, because Strato V is not a Profile 7 player.
Strato V is a Profile 5 player. There is no concept in Profile 5 of an enhancement layer, because the dynamic per-scene metadata and the extra color detail is present in the base layer. How? Profile 5 takes the same original 12-bit master image as used for Profile 7, but instead of splitting it into base and FEL enhancement layers, it converts it from YCbCr color format into ICtCp before encoding.
ICtCp is a color format that is specifically designed to represent color in a way that more closely matches how the human visual system perceives color. Basically, our visual system is more sensitive to changes in certain parts of the color/brightness spectrum, and less sensitive in others, and ICtCp takes advantage of this to make more efficient use of the bits available. With 10 bits of information, ICtCp color encoding produces an image with the same sort of imperceptible pixel-to-pixel changes that would require 12 bits to accomplish if using YCbCr. To emphasize: it’s a more efficient representation, tailor made to human perception.
By transforming the 12-bit YCbCr master image to ICtCp, Profile 5 is able to use the 10-bit video decoding engines present on current system-on-chip designs like the one used in Strato V, and it has no need for a second decoder, which these chips don’t have. Since HDMI doesn’t support ICtCp, Profile 5 players like Strato V convert the decoded ICtCp color back to 12-bit YCbCr during playback and send that 12-bit signal to the display.
The result is that Strato V’s Profile 5 playback is accomplishing the same thing as Profile 7 with FEL — it’s recovering the fidelity of the 12-bit YCbCr master image and delivering it alongside dynamic metadata. And, as a bonus, if connected to an HDR10 display, Strato V will convert that ICtCp image to an HDR10-compatible output that retains the 12-bit YCbCr. You do lose the dynamic metadata since that’s not supported by HDR10.
Obviously you can’t get something for nothing, and in this case the trade-off compared to Profile 7 is that Profile 5 does not have a backward-compatible HDR10 base layer that could work on something like Strato C. This is why there are different HDR10 and Dolby Vision downloads on the store. But by making that trade, Profile 5 is able to run on a single decoder and deliver the same benefits in terms of color fidelity that Profile 7 requires dual decoders to produce.
So, to sum up — saying “Strato V doesn’t have FEL” is not telling the full story. Both Profile 7 with FEL and Profile 5 as implemented on Strato V are doing the same thing: preserving the fidelity of the 12-bit master file. The results are not identical, because both approaches have lossy elements, but they are comparable.
I hope this helps to clarify matters.
This is going to be a bit of a technical deep dive to address a topic that has come up in several threads here on the forum recently. People who are familiar with Dolby Vision as implemented on Blu-ray Disc players may have heard the terms “FEL” and “MEL”, and these terms are now being used here when talking about Strato V. However, as I’ll explain below, these terms really don’t apply to Strato V, so I’d like to clear up any confusion.
First, four quick bits of background that are going to be important.
- In YCbCr color space with the PQ curve and BT.2020 colorimetry, it takes 12-bits of precision such that the smallest possible change between adjacent pixels is imperceptible to a human.
- Consequently, the master files that are used to create both HDR10 and Dolby Vision content use 12 bits per color component.
- Video encoding and (especially) decoding hardware is designed to deal with 10-bit information, not 12-bit. This is where the “10” in HDR10 comes from. With 10-bit YCbCr color, the smallest possible change can be visible, which is why we sometimes see banding in large areas of similar color, like a blue sky.
- It would be really great to preserve the imperceptible steps from the 12-bit master, so one of the goals of Dolby Vision was to find a way (or ways) to effectively fit 12 bits of color information into 10 bits on the video decoder.
Because Profile 7 was designed for optical discs, and there was a desire to have the discs be backward-compatible with HDR10-only players, Profile 7 starts with an HDR10 “base layer” that can be played on any 4K player. The base layer is then augmented with an “enhancement layer” when played on a Dolby Vision-capable player. (Decoding that enhancement layer is the part that requires the second video decoding engine mentioned above.) The enhancement layer in Profile 7 contains two elements:
- Dynamic per-scene metadata. Unlike HDR10’s static metadata that includes relatively crude information about the movie as a whole, Dolby Vision dynamic metadata has information for literally every shot in the film, and that metadata can be rich and extensive to allow the colorist to fine tune how each shot will be rendered to best represent the original intent.
- (Optional) “Residual” color information that, when combined with the 10-bit HDR10 base layer, reproduces the additional color information present in the original 12-bit master image.
Some folks have said things like, “Strato V doesn’t have FEL,” or have likened Strato V’s Dolby Vision implementation to Profile 7 with MEL. But this statement is not correct, because Strato V is not a Profile 7 player.
Strato V is a Profile 5 player. There is no concept in Profile 5 of an enhancement layer, because the dynamic per-scene metadata and the extra color detail is present in the base layer. How? Profile 5 takes the same original 12-bit master image as used for Profile 7, but instead of splitting it into base and FEL enhancement layers, it converts it from YCbCr color format into ICtCp before encoding.
ICtCp is a color format that is specifically designed to represent color in a way that more closely matches how the human visual system perceives color. Basically, our visual system is more sensitive to changes in certain parts of the color/brightness spectrum, and less sensitive in others, and ICtCp takes advantage of this to make more efficient use of the bits available. With 10 bits of information, ICtCp color encoding produces an image with the same sort of imperceptible pixel-to-pixel changes that would require 12 bits to accomplish if using YCbCr. To emphasize: it’s a more efficient representation, tailor made to human perception.
By transforming the 12-bit YCbCr master image to ICtCp, Profile 5 is able to use the 10-bit video decoding engines present on current system-on-chip designs like the one used in Strato V, and it has no need for a second decoder, which these chips don’t have. Since HDMI doesn’t support ICtCp, Profile 5 players like Strato V convert the decoded ICtCp color back to 12-bit YCbCr during playback and send that 12-bit signal to the display.
The result is that Strato V’s Profile 5 playback is accomplishing the same thing as Profile 7 with FEL — it’s recovering the fidelity of the 12-bit YCbCr master image and delivering it alongside dynamic metadata. And, as a bonus, if connected to an HDR10 display, Strato V will convert that ICtCp image to an HDR10-compatible output that retains the 12-bit YCbCr. You do lose the dynamic metadata since that’s not supported by HDR10.
Obviously you can’t get something for nothing, and in this case the trade-off compared to Profile 7 is that Profile 5 does not have a backward-compatible HDR10 base layer that could work on something like Strato C. This is why there are different HDR10 and Dolby Vision downloads on the store. But by making that trade, Profile 5 is able to run on a single decoder and deliver the same benefits in terms of color fidelity that Profile 7 requires dual decoders to produce.
So, to sum up — saying “Strato V doesn’t have FEL” is not telling the full story. Both Profile 7 with FEL and Profile 5 as implemented on Strato V are doing the same thing: preserving the fidelity of the 12-bit master file. The results are not identical, because both approaches have lossy elements, but they are comparable.
I hope this helps to clarify matters.