feat(server): read additional lens exif tags

[Snowknight26]

Description

exiftool provides a tag named Composite:LensID that is derived from several dozen different lens-related tags since manufacturers don't have a standard way of writing camera lens information.

In an effort to provide the best lens information possible in immich, LensID is now read, with a fallback of Composite:LensSpec, Composite:LensType and the existing LensModel.

Out of about 85k RAW files that I had access to, exiftool provided a Composite:LensID tag for all of them, though few with questionable values.

This is a drastic improvement over the current approach, which, with the same sample of RAW images, read only about 8% of the lens information.

LensID distribution

LensID	Count
20.7mm	120
26.3mm f/2.2	7
Canon EF 100mm f/2.8L Macro IS USM	11
Canon EF 24-70mm f/2.8L USM	24
Canon EF 40mm f/2.8 STM	3
Canon EF 50mm f/1.8 STM	42
Canon EF-S 18-135mm f/3.5-5.6 IS STM	263
Canon EF-S 18-55mm f/3.5-5.6 IS II	58
DJI MFT 15mm F1.7 ASPH	997
E 70-350mm F4.5-6.3 G OSS	25
FE 24-70mm F2.8 GM II	12
LEICA D VARIO-ELMARIT 14-50mm	1359
LEICA DG 100-400mm F4.0-6.3	18
LEICA DG 10-25mm F1.7	738
LEICA DG 12-60mm F2.8-4.0	239
LEICA DG 50-200mm F2.8-4.0	95
LEICA DG 8-18mm F2.8-4.0	194
Leica DG Vario-Elmar 100-400mm F4.0-6.3 Asph. Power OIS	4094
Leica DG Vario-Elmarit 12-60mm F2.8-4 Asph. Power OIS	8641
Leica DG Vario-Elmarit 50-200mm F2.8-4 Asph. Power OIS	10432
Leica DG Vario-Elmarit 8-18mm F2.8-4 Asph.	6401
Leica DG Vario-Summilux 10-25mm F1.7 Asph.	22346
LUMIX G 14mm F2.5	648
LUMIX G 20mm F1.7	5280
LUMIX G 25mm F1.7	4
Lumix G 25mm F1.7 Asph.	171
LUMIX G 42.5mm F1.7	739
Lumix G 42.5mm F1.7 Asph. Power OIS	2086
LUMIX G MACRO 30mm F2.8	11
Lumix G Macro 30mm F2.8 Asph. Mega OIS	292
LUMIX G VARIO 100-300mm F4.0-5.6	3
Lumix G Vario 100-300mm F4.0-5.6 Mega OIS	75
OLYMPUS M.75mm F1.8	906
Olympus M.Zuiko Digital ED 75mm F1.8	1208
Sony E 18-135mm F3.5-5.6 OSS	561
Sony E 70-350mm F4.5-6.3 G OSS	2303
Sony FE 200-600mm F5.6-6.3 G OSS + 1.4X Teleconverter	426
Sony FE 24-70mm F2.8 GM II	2791
Sony FE 24-70mm F4 ZA OSS	44
Sony FE 35mm F2.8 ZA	3
Sony FE 70-200mm F2.8 GM OSS II	62
Sony FE 90mm F2.8 Macro G OSS	993
Tamron 28-75mm F2.8 Di III RXD	2
Tamron SP AF 28-75mm f/2.8 XR Di LD Aspherical [IF] Macro	80
Unknown (0 ff ff)	21
Unknown (8 01 10)	11512
Voigtlander NOKTON 50mm F1 Aspherical	77
Zeiss Batis 25mm F2	84
Zeiss Batis 85mm F1.8	6

LensType distribution

LensSpec	Count
	83842
Leica DG Vario-Elmarit 12-60mm F2.8-4 Asph. Power OIS	1738
Leica DG Vario-Elmarit 8-18mm F2.8-4 Asph.	916
Lumix G 42.5mm F1.7 Asph. Power OIS	9
Unknown (8 01 10)	2

LensSpec distribution

N/A, didn't have any

LensModel distribution

LensModel	Count
	78644
----	77
28-75mm	80
E 18-135mm F3.5-5.6 OSS	561
E 28-75mm F2.8-2.8	2
E 70-350mm F4.5-6.3 G OSS	2328
EF100mm f/2.8L Macro IS USM	11
EF24-70mm f/2.8L USM	24
EF40mm f/2.8 STM	3
EF50mm f/1.8 STM	42
EF-S18-135mm f/3.5-5.6 IS STM	263
EF-S18-55mm f/3.5-5.6 IS II	51
FE 200-600mm F5.6-6.3 G OSS + 1.4X Teleconverter	426
FE 24-70mm F2.8 GM II	2803
FE 24-70mm F4 ZA OSS	44
FE 35mm F2.8 ZA	3
FE 70-200mm F2.8 GM OSS II	62
FE 90mm F2.8 Macro G OSS	993
ZEISS Batis 1.8/85	6
ZEISS Batis 2/25	84

There are other LensID examples that are less than idea, such as the example in #17050, but I think that's better than not reading the data at all.

Various findings:

• Sony, Canon populate LensModel
• Panasonic populates LensType
• Nikon populates LensSpec?

Implements #17050

How Has This Been Tested?

• Refreshed metadata for an asset with each tag

Screenshots (if appropriate)

Checklist:

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation if applicable
• I have no unrelated changes in the PR.
• I have confirmed that any new dependencies are strictly necessary.
• I have written tests for new code (if applicable)
• I have followed naming conventions/patterns in the surrounding code
• All code in src/services/ uses repositories implementations for database calls, filesystem operations, etc.
• All code in src/repositories/ is pretty basic/simple and does not have any immich specific logic (that belongs in src/services/)

[Snowknight26]

Couple outstanding questions based on some example LensID values I've seen:

1. Are we ok with values like the following:
   1. ----
   2. Unknown (0 ff ff)
   3. Unknown (8 01 10)
   4. Unknown (E1 40 19 36 2C 35 DF 0E) Tamron 10-24mm f/3.5-4.5 Di II VC HLD (B023) ?
   5. Unknown 6-30mm
   6. Sony FE 100mm F2.8 STF GM OSS or Sony FE 70-200mm F2.8 GM OSS or Sony FE 70-200mm F2.8 GM OSS + 1.4X Teleconverter or Sony FE 70-200mm F2.8 GM OSS + 2X Teleconverter or Sigma 70-200mm F2.8 DG OS HSM | S

If yes.. cool, nothing to do.
If no:

2. Should we consolidate ---- and Unknown (*) into a single value (thinking null)?
   • This covers 1.i, 1.ii and 1.iii
3. Should we consolidate ---- and Unknown* into a single value (thinking null)?
   • This covers 1.i, 1.ii, 1.iii, 1.iv, and 1.v
   • Might be a loss of information for 1.iv
4. Should we consolidate ---- and Unknown (*)* into a single value (thinking null)?
   • This covers 1.i,1.ii, 1.iii and 1.iv
   • Might be a loss of information for 1.iv
5. Should we attempt to infer the lens from a value like Unknown (E1 40 19 36 2C 35 DF 0E) Tamron 10-24mm f/3.5-4.5 Di II VC HLD (B023) ? as Tamron 10-24mm f/3.5-4.5 Di II VC HLD (B023)?
   • This covers 1.iv

[zackpollard]

Couple outstanding questions based on some example LensID values I've seen:

1. Are we ok with values like the following:
   
   1. `----`
   2. `Unknown (0 ff ff)`
   3. `Unknown (8 01 10)`
   4. `Unknown (E1 40 19 36 2C 35 DF 0E) Tamron 10-24mm f/3.5-4.5 Di II VC HLD (B023) ?`
   5. `Unknown 6-30mm`
   6. `Sony FE 100mm F2.8 STF GM OSS or Sony FE 70-200mm F2.8 GM OSS or Sony FE 70-200mm F2.8 GM OSS + 1.4X Teleconverter or Sony FE 70-200mm F2.8 GM OSS + 2X Teleconverter or Sigma 70-200mm F2.8 DG OS HSM | S`

I personally think all the unknown and ---- values should be ignored entirely, we shouldn't have those come through to the application. As for vi, it seems pretty useless too, it's giving like 4 different possible lenses. Aside from that, the values in the rest of the list seem much better.

[Snowknight26]

I personally think all the unknown and ---- values should be ignored entirely, we shouldn't have those come through to the application. As for vi, it seems pretty useless too, it's giving like 4 different possible lenses. Aside from that, the values in the rest of the list seem much better.

Didn't feel comfortable adding code for that last example without the risk of false positives but I've now made the others work the same way as if the exif info wasn't set.