Blackmagic Raw vs 10-bit and 8-bit Log and Rec.709 (Video)

[Horshack]

Nikon recently announced support for Blackmagic raw on their Z cameras, recordable over HDMI. I don't have the raw upgrade on my Z6 but out of curiosity of raw video I recently purchased a Blackmagic Pocket Cinema 4K, which supports Blackmagic raw internally and over USB-C to an SSD.

After learning a basic raw grading workflow (Davinci Resolve) I started doing some comparisons to see if raw is worth the trouble. I imagine others might be curious as well so I'm sharing my initial test.

I shot a short raw video clip. I then generated various Log and Rec. 709 versions of that same clip so that I have identical source and scene material to compare. I generated these versions in Davinci Resolve using a straight conversion, exporting to H.265 in combinations of 10 and 8 bit Log (V-Log) and Rec. 709, all with data rates around 100Mbps.

I then imported all the clips, along side the original raw clip, and did a single color grade to all: increasing Davinci's contrast slider from 1.0 to 1.49. I then compared a 100% crop of the sky from each clip to show any banding/posterization. I also included screen grabs of the scopes and histograms to demonstrate the degree of posterization from the contrast adjustment.

This is my first attempt at this kind of comparison so there is a greater-than-even chance I screwed something up in the methodology

Results

• Raw: No posterization visible at 100% or 200%. No posterization on scopes.
• V-Log 10-bit: No posterization visible at 100% or 200%. Small amount of posterization visible on scopes.
• V-Log 8-bit: Worse of all clips. Posterization easily visible at 100%. Severe posterization visible on scope. I assume this worst-showing is the result of insufficient bit depth to support the additional tonal workload of the compressed log curve
• Rec. 709 10-bit. No posterization visible at 100% or 200%. No posterization on scopes.
• Rec. 709 8-bit. Tiny amount of posterization visible at 100% and 200%. Moderate amount of posterization on scopes.

I plan to perform more comparisons, using additional grading adjustments to see what adjustments cause what to "break" in terms of IQ.

Here is a direct link to the image below: Full-size image


Comparison of Blackmagic Raw vs Log vs Rec. 709 at 10-bit and 8-bit, after a single grading adjustment

 

[Joop_S]

On 2-25 19:00-19:30 is Blackｍagic Design on the Nikon online CP+ site with:

An introduction to cinema starting with Blackmagic Design and Z

 

[plumberdp69]

This is my first attempt at this kind of comparison so there is a greater-than-even chance I screwed something up in the methodology

Results

• Raw: No posterization visible at 100% or 200%. No posterization on scopes.
• V-Log 10-bit: No posterization visible at 100% or 200%. Small amount of posterization visible on scopes.
• V-Log 8-bit: Worse of all clips. Posterization easily visible at 100%. Severe posterization visible on scope. I assume this worst-showing is the result of insufficient bit depth to support the additional tonal workload of the compressed log curve
• Rec. 709 10-bit. No posterization visible at 100% or 200%. No posterization on scopes.
• Rec. 709 8-bit. Tiny amount of posterization visible at 100% and 200%. Moderate amount of posterization on scopes.

I plan to perform more comparisons, using additional grading adjustments to see what adjustments cause what to "break" in terms of IQ.

Here is a direct link to the image below: Full-size image

Many thanks! That is what I have been wondering too on whether to venture into 10 bit etc.

I have not bought the extra gears and will follow you on the results before getting out my wallet.

Cheerio.

 

[plumberdp69]

On 2-25 19:00-19:30 is Blackｍagic Design on the Nikon online CP+ site with:

An introduction to cinema starting with Blackmagic Design and Z

Thanks.

Tried to register for that (assuming it is a webinar), cant find the web page. Hope that they will have that video for wider viewing later.

 

[James Stirling]

Nikon recently announced support for Blackmagic raw on their Z cameras, recordable over HDMI. I don't have the raw upgrade on my Z6 but out of curiosity of raw video I recently purchased a Blackmagic Pocket Cinema 4K, which supports Blackmagic raw internally and over USB-C to an SSD.

After learning a basic raw grading workflow (Davinci Resolve) I started doing some comparisons to see if raw is worth the trouble. I imagine others might be curious as well so I'm sharing my initial test.

I shot a short raw video clip. I then generated various Log and Rec. 709 versions of that same clip so that I have identical source and scene material to compare. I generated these versions in Davinci Resolve using a straight conversion, exporting to H.265 in combinations of 10 and 8 bit Log (V-Log) and Rec. 709, all with data rates around 100Mbps.

I then imported all the clips, along side the original raw clip, and did a single color grade to all: increasing Davinci's contrast slider from 1.0 to 1.49. I then compared a 100% crop of the sky from each clip to show any banding/posterization. I also included screen grabs of the scopes and histograms to demonstrate the degree of posterization from the contrast adjustment.

This is my first attempt at this kind of comparison so there is a greater-than-even chance I screwed something up in the methodology

Results

• Raw: No posterization visible at 100% or 200%. No posterization on scopes.
• V-Log 10-bit: No posterization visible at 100% or 200%. Small amount of posterization visible on scopes.
• V-Log 8-bit: Worse of all clips. Posterization easily visible at 100%. Severe posterization visible on scope. I assume this worst-showing is the result of insufficient bit depth to support the additional tonal workload of the compressed log curve
• Rec. 709 10-bit. No posterization visible at 100% or 200%. No posterization on scopes.
• Rec. 709 8-bit. Tiny amount of posterization visible at 100% and 200%. Moderate amount of posterization on scopes.

I plan to perform more comparisons, using additional grading adjustments to see what adjustments cause what to "break" in terms of IQ.

Here is a direct link to the image below: Full-size image


Comparison of Blackmagic Raw vs Log vs Rec. 709 at 10-bit and 8-bit, after a single grading adjustment

Thank you for taking the time to do this , it is much appreciated

--
Jim Stirling:
It is undesirable to believe a proposition when there is no ground whatever for supposing it true” Russell
Feel free to tinker with any photos I post

 

[beatboxa]

Nikon recently announced support for Blackmagic raw on their Z cameras, recordable over HDMI. I don't have the raw upgrade on my Z6 but out of curiosity of raw video I recently purchased a Blackmagic Pocket Cinema 4K, which supports Blackmagic raw internally and over USB-C to an SSD.

After learning a basic raw grading workflow (Davinci Resolve) I started doing some comparisons to see if raw is worth the trouble. I imagine others might be curious as well so I'm sharing my initial test.

I shot a short raw video clip. I then generated various Log and Rec. 709 versions of that same clip so that I have identical source and scene material to compare. I generated these versions in Davinci Resolve using a straight conversion, exporting to H.265 in combinations of 10 and 8 bit Log (V-Log) and Rec. 709, all with data rates around 100Mbps.

I then imported all the clips, along side the original raw clip, and did a single color grade to all: increasing Davinci's contrast slider from 1.0 to 1.49. I then compared a 100% crop of the sky from each clip to show any banding/posterization. I also included screen grabs of the scopes and histograms to demonstrate the degree of posterization from the contrast adjustment.

This is my first attempt at this kind of comparison so there is a greater-than-even chance I screwed something up in the methodology

Results

• Raw: No posterization visible at 100% or 200%. No posterization on scopes.
• V-Log 10-bit: No posterization visible at 100% or 200%. Small amount of posterization visible on scopes.
• V-Log 8-bit: Worse of all clips. Posterization easily visible at 100%. Severe posterization visible on scope. I assume this worst-showing is the result of insufficient bit depth to support the additional tonal workload of the compressed log curve
• Rec. 709 10-bit. No posterization visible at 100% or 200%. No posterization on scopes.
• Rec. 709 8-bit. Tiny amount of posterization visible at 100% and 200%. Moderate amount of posterization on scopes.

I plan to perform more comparisons, using additional grading adjustments to see what adjustments cause what to "break" in terms of IQ.

Here is a direct link to the image below: Full-size image

Good stuff, and it sounds like the results align to what would be expected.