A workflow for using video frames for observations

I have been experimenting with a new way of getting iNaturalist observations. I ride my bike at night on a paved path through good habitat and see a lot of nocturnal wildlife. I’ve been making observations for a long time while on these rides. When I see an animal I want to observe, I stop the bike, open my camera app, then start taking pictures. The problem with this methodology, is the animals often run or fly off before I can start photographing. So as the experiment, I bought a GoPro (Hero12 black model). I mount it on my wrist with their standard wrist mount. I run GeoTagPhotosPro to gather location information. I start recording in video mode at the start of the ride, then whenever I see something I want to observe, I immediately point my wrist at it. I nearly always slow or break during this process and continue pointing the camera at the subject. Then I stop and restart the recording, so each video observation is at the end of a video. After a night of riding I have a series of videos, one for each observation. My rides last around 2 hours, so I can get a lot of observations.

Afterwards, comes the processing stage, which I find more time consuming and error-prone than I’d like. This is where the question comes in. How can I do the following better: 1) I transfer all of the videos to the GoPro Quick app on my iPhone. Then for each video, the following: 2) In Quick, trim each video, so it shows only the section with the animal. 3) Save it as a new video. 4) Export it to photos. 5) Load the video into the iPhone photos app, 6) Select “Adjust Date & Time” and push “Revert” (this is a key step as at this point the timestamp is when the original video started recording. For some reason the time of the start of the trimmed video is stored in “Original”), 7) crop the video (extreme amount of cropping is needed due to GoPros wide angle lens), 8) save the edited video, 9) load the video into the iPhone FrameGrabber app, 10) export 1 or more frames (these frames will all have the timestamp of the video). 11) Do more cropping of these photos using the iPhone photos app. At this point, the photos are ready for geotagging and loading as observations.

I’m generally quite happy the end product, it’s just the tedium involved in getting them. Here are some armadillos I observed on the last ride: https://www.inaturalist.org/observations/identify?reviewed=any&quality_grade=needs_id%2Cresearch&verifiable=any&user_id=dan_johnson&d1=2023-10-19&d2=2023-10-21&taxon_id=47075&place_id=any. The photo quality is not great, but easily good enough to be identifiable.


Which such a workflow the first thing I would do is change to a laptop or desktop instead of a smartphone. Generally, these devices speed up the workflow just by their processing speed and ability to multi task.


I called the help line for GoPro. They said they stopped Windows support because most people want to use Mobile devices.

1 Like

I take videos frequently for similar reasons, though usually it’s while in a vehicle or on foot. I record directly on my phone so there’s already a skipped over step because it’s on the phone immediately. Then I go frame by frame and take screenshots. If I need to edit a picture to make it clearer or brighter, I edit the screenshots after.

I also have to manually add the location and date, which I just do on the iPhone’s photo app.

That being said, sometimes an opportunity is just lost and there’s not much to take from videos. Your system seems like there’s a lot more steps but probably yields better results.

1 Like

is your end product just a photo, or are you trying to get a photo and a cropped video? if you want just a photo, there are lots of ways to export frames from videos.

my understanding is that all GoPros produce mp4s in either H.264 or H.265. so just about any video editor should handle them. i’m not sure why you necessarily need the Quik app, although you may be interested to know that a new desktop version is supposed to be available for the Mac in November and the middle of 2024 for Windows… if you are a GoPro subscriber.

For now, the main reason for using GoPro Quik is it efficiently handles huge videos. I tried using a couple other editors and they are so slow in handling them as to be useless. I don’t remember which, but for one editor, loading appeared to take about 1 hour. Quik is instantaneous. If the frames are identifiable, that’s the end goal. If not, video can be identifiable, so I’ll make GIFs, often in slow motion. Also I sometimes extract sounds. I did for observations this morning from last nights ride here: https://www.inaturalist.org/observations/identify?reviewed=any&quality_grade=needs_id%2Cresearch&iconic_taxa=Mammalia&verifiable=any&user_id=dan_johnson&created_on=2023-10-28&createdDateType=exact&place_id=any

1 Like

hmmm… as long as you’re not trying to load giant files directly into the timeline of a video editor and/or converting a lot of stuff upfront, it shouldn’t take that much time to load.

a lot of video editors have something where you can view and mark which parts of the source videos you want before you drop them into the timeline, and you should be able to do your splitting and even export video directly from here. once you have just the chunk that you want, you can go through your clip in the timeline to pull out the frames you like, too. (i don’t have access to my Windows machine for a week or so, but i believe even a free program like Shotcut should handle your workflow as long as you split your giant video in that source viewer before trying to do other stuff to it.)

or if you don’t even need to generate an actual video clip, you can export frames from some video players – even one that’s free to use (ex. VLC).

1 Like

Is it even possible to add a video on the site?

no. many folks have requested this functionality, but the infrastructure (and cost) needed to process and host a lot of video probably is beyond what a small organization like iNaturalist is realistically capable of taking on. some folks post animated GIFs, but this is not true video (which would include sound), doesn’t display correctly in all iNat interfaces, and is frowned upon though not punished by the iNat overseers.

for now, you can post your videos elsewhere and just link to those in your observation, like i’ve done in the description of this observation: https://www.inaturalist.org/observations/157901323.

1 Like

I may be slightly off from your question but I use a different method which does save time with taking videos - similar to @pisum

I take a video on my phone and upload to YouTube (I have a playlist titled iNaturalist Observations). I then make an iNats observation on my phone - this gives time and location (otherwise I forget). Generally the quality of the image is not good or the creature has disappeared. I make a note that more detail to come later. Later on the pc, I copy and paste the link into the notes on iNats and on YouTube.

This is an example of one of my observations - https://www.inaturalist.org/observations/145412208 Although after a very frustrating time of failure, I did manage to get a good image!

A bonus of having the video link is that there is observation of the creature’s behaviour.

1 Like

Does that model have a time lapse photo mode? My ancient Hero4 does. I can set it to take a photo every second, for example. I can also change the FOV to take it out of wide angle mode, which means less cropping. That would probably get you better results (less blur, higher resolution) if photos are the end result you’re looking for.

If you prefer to pull stills from video, you should consider a way to identify and export the keyframes. They will have less blur than other frames. Some video players allow you to skip directly between keyframes and export the ones you want. If you’re tech savvy, you can use ffmpeg to export all the key frames. That would leave you with a bunch of stills similar to what you’d get from the time lapse mode.

1 Like

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.