Replies: 4 comments
-
I assume there is a bug in immich. Normally for pictures the exif entry datetimeoriginal is used for sorting. This entry is stored only in local time. There is no information of timezone. Many cameras have no gps and so the offset time is not known. Older pictures has no gps data and the only information of the taken time is the local time stored in the exif data datetimeoriginal. For MP4 the createdate can only be stored in UTC time. The server software has to calculate therefore the time difference that the timeline between photos and videos is correct. Otherwise the timeline were a total mess. I have found out actually that there is a time difference between photos with gps data and without, because the offset time is somehow calculated. In my opinion this is wrong and only the datetimeoriginal should be used for the local time. So Windows and other programs (plex, Google.......) handle the data |
Beta Was this translation helpful? Give feedback.
-
@MMauro94 and @DeltaTango69 I started a Discord focus discussion on this point here: https://discord.com/channels/979116623879368755/1206309678578016277/1206309678578016277, because I agree with your points. Your input would be super appreciated. |
Beta Was this translation helpful? Give feedback.
-
Yes, I believe it is a bug. I have hundreds of photos that are are sorted and presented as if they were taken 1 or 2 hours later than they actually were. Is it trying to add timezone or something? What is curious that it doesn't happen with photos taken by a smartphone. These are fine. If I didn't have some albums where both cameras and smartphone photos were mixed together, I would not even care much. I have 2 photos, one from smartphone and other from a camera taken, say, at 14:04. Everything seems to be correct in EXIF data of the one taken by a camera and still, Immich says it was taken at 16:04 GMT + 02:00. The smartphone displays 14:05 GMT + 02:00, all correct. I can even provide some exiftool output to prove it. When I download both photos into my iPhone directly from Immich, they both show 14:04. It is a mystery for me. |
Beta Was this translation helpful? Give feedback.
-
I'm working on resolving the issues with timezone detection for DSLR pictures. At least on my Sony A7 first gen it's possible to infer the timezone from the exif metadata. To fix this issue for other camera's as well, I need some example pictures. To submit example pictures, please follow the following steps:
Upload both pictures to the following github issue: photostructure/exiftool-vendored.js#187. Please document clearly which image is which timezone and which camera you used. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hello everyone! First of all thanks for all the effort you put into this software, it's really great. I'd like to show you a problem I have and propose a possible solution.
The Problem
I'm often in this situation:
The problem I'm facing is that the first photos are completely out of order wrt the others. This is because the default timezone of the server is being used for the first ones, while the actual timezone derived from gps data is used for the others.
Setting the "correct" timezone for the server, however, is very problematic. In my case:
The solution
Option 1
My proposed solution is to split the
exif.dateTimeOriginal
DB field in two:dateTime
will be of the postgres typetimestamp
: this contains just the date and time, without the timezone (basically aLocalDateTime
for those familiar with Joda time/Java APIs)timezone
: just the timezone. This could be an offset (int) or even aZoneId
likeEurope/Rome
, etc.This allows, in my opinion, for a lot of things to "fall into place":
Option 2
Another option to fix the same problem would be to fix the heuristics of the guessing of the timezone based on photos shot close in time. For instance, if I have these photos:
We can safely assume that the user didn't teleport to another country in the span of 10 minutes and extrapolate the TZ for photos A and D.
I'm just mentioning this option as a possibility, but I don't really like it because:
Option 3
Last thing that comes to mind is for each user to be able to set in which timezone they've been in for certain periods. For people who don't travel super often, this might be an easy thing to setup:
For most people this would be a very short table with one home timezone and a few entries for each one of their travels in a different TZ.
This way whenever a timezone is needed, this information can be looked up easily.
This could also be automagically assumed based on aggregated geo information of the library, similarly to what's described in Option 2.
My personal preference is Option 1, as I find it the best way to deal with this kind of things.
Of course, things could also be mixed. For instance, separate the fields as in option 1 but use the lookup table in option 3.
I'm curious to hear your thoughts about this and hopefully reach a consensus on what's the best course of action :)
Thanks,
Mauro.
PS: slightly unrelated, I have another (easier) sorting problem with scanned analog photos: since I don't know exactly the day/time in which they where shot, I've just put a date like
1994-10-10T10:00:00Z
. It would be nice to order based on the original filename if the order date/time are identical.Beta Was this translation helpful? Give feedback.
All reactions