Is it better to use "real" 10 Bit or 14 BitLess

[hamanohashidate]

As far I read on internet is choosing 14 bitsless means the color depth is reduced, which could imply that the number of brightness levels per channel is less than 16,384 14 bits (but potentially more than 8 bits)

So as I understand no garanty that I will have more than 10 bits ...

So is it better to use a "real" 10 bit int the configuration and be sure that I will have 10 bit or in magic lantern 14 bitsless is better ?

For info I have a 5dm3 in my situation

I think the question is like to compare is it better to use mp3 compressed in 256kbps variable bitrate (VBR) (so no guarantee to get always more than 192... it can be sometime 56 or 128 ! ) or use a real 192 kbps compression ? In mp3 my answer is use 192 kbps real compression than a 256 VBR ..but what is your answer in bit video comparaison ?

Let me know if it the same or not. I would like to understand.

[Walter Schulz]

I think there is some basic misunderstanding and mixup of information bits on your side.

Some facts:

- Lossless compression is *lossless*. After decompression there is no difference between the data without/before and with compression.
It works in the same way (but with a different algorithm) as ZIP: You can zip and unzip a data file as often as you wish and data after decompression is always identical to original data before.
Therefore it makes no sense to compare "14 bit lossless" to "10 bit". You are comparing 14 and 10 bit data aquisition anyway.

- Indeed the highest value stored in 14 bit RAW/MLV may not be equal to 16383. And actual range between lowest and highest value in a given RAW may be lower that "0 to highest value". Mainly for two reasons:
-- Underexposure. Well, kind of. There may be some "hot pixel" having a significant higher level than data actually representing light captured by a photosite.
But indeed you are actually not using the full bit capacity when underexposed. Underexposure by 1 EV reduces bit depth by 2 -> "14 bit" is actually using 13 bit. With some hot pixel/noise pixels reaching higher levels.
-- "White level" and "Black level subtraction":
White Level: This value is the highest bit value a RAW may contain. And it may not be 16383 but significant lower. But it is higher than 2^13 - 1 = 8191. The actual white level may depend on sensor type/generation and ISO! For this and a short explanation what is actually happing see https://www.rawdigger.com/node/179

I will not dive into MP3 compression because IMO it doesn't make sense to discuss it in this context.

Please come back if there a further questions.

[hamanohashidate]

thank you for your clarification