-
-
Notifications
You must be signed in to change notification settings - Fork 2.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Exif date/time isn't parsed correctly #9331
Comments
I'm seeing similar problems on my GrapheneOS phone as well. Seems to impact videos and photos differently. Sometimes tagged with local time zone, other times, UTC. -- but odd, because it's not just the wrong time zone, it can be the wrong day in UTC. IE videos I took last night are saying they're in the future -- for tonight. |
I see similar issues on my end. In my case I my pictures where saved into iCloud Photos from a DJI Osmo 3, and when the app imported them into Immich the datetime was wrong by 2 hours. I removed all pictures and tried to import them manually via Attaching one of the pictures that has that issue. |
Yeah most of my issues result from my Osmo Pocket 3 as well. My guess is this is because the photos lack GPS data, but it would be nice if it still had accurate time.
One thing I considered is that I tend to rename the lrf files to MP4 so I can upload those, and I keep the full size media elsewhere for editing. Perhaps DJI does something weird with the LRF exif.
…-------- Original Message --------
On 5/12/24 2:31 AM, Felipe Martin wrote:
I see similar issues on my end. In my case I my pictures where saved into iCloud Photos from a DJI Osmo 3, and when the app imported them into Immich the datetime was wrong by 2 hours.
I removed all pictures and tried to import them manually via
the web interface, same result.
Attaching one of the pictures that has that issue.
[DJI_20240511164405_0094_D.JPG](https://github.com/immich-app/immich/assets/812088/00b825d3-c5c5-433c-9f49-1954bfe2b007)
—
Reply to this email directly, [view it on GitHub](#9331 (comment)), or [unsubscribe](https://github.com/notifications/unsubscribe-auth/ADS37XGWX2MKMKSP6ADBLC3ZB4ED3AVCNFSM6AAAAABHM2D23KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMBWGEZTONJXGY).
You are receiving this because you commented.Message ID: ***@***.***>
|
Anyone from the project that can comment back? |
Chiming in; same behavior here with older photos shot on a Motorola Moto G (1. Gen), additionaly it dates it far back (photo from 2016, Immich dates it as 2002) EDIT: Okay, I just learned that Windows Explorer doesn't always read EXIF data. |
2 weeks later: "Anyone from the project that can comment back?" |
Are you on the latest version of the mobile app/server? |
Was at the time of reporting, yes .. |
I am pretty sure the latest version was v1.105.0 two weeks ago. Can you help upgrade it and test it again? Also include the troubled file will help with troubleshooting as well. |
I just updated to v1.105.1 and refreshed the metadata for my entire library, still having the same issue. Sample file in my previous comment. |
Just updated to
Tested and it is still borked .. |
Attached sample file ... That becomes like follows in Immich ... |
Almost 4 weeks later: "Anyone from the project that can comment back?" |
@ygaeon The timezone is parsed based on a few heuristics, so it isn't easy to explain precisely which one is used. From my observation in the past, CalyxOS strips GPS data, which the parsing mechanism uses to identify the correct timezone You can read more here https://photostructure.github.io/exiftool-vendored.js/#md:dates |
Thanks @alextran1502 and yes, CalyxOS do but I'm using "OpenCamera" which does not strip GPS, afaik. Though, I fail to see why GPS would be needed in this case as the exif data is present. Please just use it when it is there?
|
I checked the attached file, following your link @alextran1502 ... Heuristic 1: explicit metadata
Heuristic 2: GPS locationNo GPS location available Heuristic 3: UTC timestamps
.. so I guess that exiftool-vendored would assume UTC in this case .. making all photos off by the 1-2 hours I see .. In this case, would it not be possible to "attach" the timezone (from the mobile device) to the image/video being uploaded to the store? |
Opened this over 4 months ago, still have the same issues. Anyone more knowledgeable have any comments or feedback? Server Version: v1.114.0 |
Some observations Starting info:
The way I observe immich working when it does not have any information related to timezone:
What would likely be an expected behavior is this:
|
Many cameras without GPS just store DateTimeOriginal tags without timezone information. We often take pictures while on vacation in timezones that are not the same as the immich server and not in UTC. I don't think immich should ever assume what the timezone is. If it can't be determined from the image metadata, then the timezone should be explicitly stated by the user during import. This is only necessary because immich likes to attach a timezone to timestamps it stores. Other photo management systems do not care about timezone and the assumption is that the DateTimeOriginal is simply in whatever timezone where the photo was captured. |
Correction to my previous comment - by timezone of the server, I actually meant the timezone the user sets using a env variable in the compose. Also, I dont think that there is a solution without having a timezone attached to an image. I mean the entire concept of time is relative. So how would any application be able to provide a timeline view of images that can be taken in multiple timezones. Everything needs to be brought to a common reference, which is UTC, and we need the timezone to do it. |
I agree it is nice to have the proper timezone attached to an image, but it was not deemed necessary when EXIF metadata was invented for use with digital cameras. I've been using the photo management site flickr for 20 years with images that don't have timezone information and the timeline features work just fine when you understand the time displayed is wherever the photo was captured. Flickr and other photo sites do not assume a timezone. Only the photographer knows the correct timezone in these cases so we need a way to pass that to immich during photo import if immich insists on storing timezone. If the photographer travels often, it can't be a fixed offset set from an env variable. |
That is confusing. Maybe getting a bit off topic here - I still do not think flicker's timeline would work well ( it can still work. Bit is the timeline accurate - dont think so). What happen's if you upload 2 images. Which image would come first in the timeline in Flicker? Flicker would need to assume a timezone for the images that dont have timezone OR just have B (11 am) come after A (10 am) just because of the time, which could most certainly be wrong. Coming back to the original problem :), Immich does have environment variable that set the timezone of images that do not have a timezone. but it might also be worthwhile to be able to edit the timezone in the GUI i think. |
It is a known drawback of EXIF that until recently it did not provide a way for cameras to record time-zone information. Any long-lived photo management system has had to deal with this. Flickr and others simply do not use timezone information for the timeline since historically it has not been available. Digital cameras existed for 20 years before geotagging became common and before the EXIF spec was updated to allow timezone with OffsetTimeOriginal. If you upload two photos to flickr, one with timezone data and one without, it will ignore the timezone and make no assumptions about timezone for the other. The timeline will still be correct and ordered exactly as captured. It would only be a problem if your timeline somehow contains photos captured by multiple photographers shooting simultaneously in different timezones. |
Thanks for some activity in this issue .. :-)
Well, if the timezone is missing in the metadata I think the closest truth would be the date and time and timezone set on the mobile device .. I mean, we usually update the time on our mobile device when we travel, right? |
I import photos when I get home, not when I'm away in a different time zone. So mobile device timezone can't be used to set image metadata during import. Mobile devices really should be smart enough to set image metadata correctly already during image capture. |
Maybe a section in the UI showing the latest import along files already stored with the option to modify them in batch, to check if everything is set ok or modify datetime settings accordingly. |
Issues with parsing date/time incorrectly should go to #12650, potential feature requests (bulk modify assets) should be a feature request (pretty sure there already is one anyways) |
The bug
(Note this might be related to / same as #4898)
We have two different models of phones,
Both cameras sync to Immich server but our photos are not in order and I started investigating.
My photos are saved with wrong date/timestamp.
I'm currently in UTC+2 (Stockholm/Europe).
Comparing EXIFs (exiftool 12.76):
Photos
Fairphone 4:
This Date/Time is translated to
Wrong time, wrong timezone ...
Samsung Galaxy Note 10:
This Date/Time is translated to
Correct time, wrong timezone ...
Videos
Fairphone 4:
This Date/Time is translated to
Which kind of translates to "correct" even if the correct should be "05/04/2024, 04:44 PM (16:44), Europe/Stockholm (+02:00)"
Samsung Galaxy Note 10:
This Date/Time is translated to
Correct time, wrong timezone ...
From https://web.archive.org/web/20180921145139if_/http://www.cipa.jp:80/std/documents/e/DC-010-2017_E.pdf one can read:
From https://en.wikipedia.org/wiki/ISO_8601 on can read:
<time>Z <time>±hh:mm <time>±hhmm <time>±hh
Now, this is clearly not what's happening in Fairphone 4 case above:
Something is clearly not being parsed correctly somewhere .. and I have thousands of photos that are sorted wrongly in the archive now ..
Please advice ..
The OS that Immich Server is running on
Debian
Version of Immich Server
v1.99.0
Version of Immich Mobile App
1.100.0 build.130
Platform with the issue
Your docker-compose.yml content
Your .env content
Reproduction steps
Relevant log output
No response
Additional information
No response
The text was updated successfully, but these errors were encountered: