How to convert a video so it can be posted on iNat?

My apologies if this has been addressed before - I just encountered it today. A previous post on the forum stated that it is important to post a video when posting observations of ants because the way they walk helps to ID them. So, yesterday I tried to post a video from my android phone and iNat said it was the wrong format. It is an mp4 and should be converted to an mp3 to post, but then it would be audio only and no help in IDing. Can anyone tell me how to convert my phone video to a video that iNat will accept? Also, for the ant people, is it true that video is important for IDing?

3 Likes

You can’t post actual videos to iNat, but you can upload a GIF (you can look up free GIF creators online).

Edit to add: I don’t think the iNat apps support GIFs, so you’ll have to upload from the website.

There is a feature request (Upload Videos) to allow users to upload videos to iNat directly, but there are some logistic hurtles that make implementing that feature hard.

5 Likes

Another workaround is to upload the video to YouTube or another video sharing site then post the link in the description of the observation

5 Likes

see Animated Observations and this post
https://www.inaturalist.org/projects/animated-observations/journal/3544-please-add-comments

The iNat staff have said they’ll likely stop supporting gifs in the future, and will convert them all to still images at that time, so your best option is probably to just take screenshots/stills of your video and host/link to the video somewhere else that’s relatively stable.

8 Likes

Or the helpfile https://help.inaturalist.org/en/support/solutions/articles/151000169940-can-i-upload-videos-to-inaturalist-

i load videos to YouTube and then put a link to them in the descriptions of associated iNat observations. examples:

a single photo might have more pixels than a typical video frame, but a whole video provides more angles, gives a better understanding of movement, and includes sounds, too. sometimes there are things that just can’t be captured as well in a photo or even a series of photos.

8 Likes

A video provides so much more information for the human eye than most any still picture! Even 3-6 seconds of video can convey so much more than a series of high-def still photos (gait, flight pattern, even body configuration ).

Sometimes when feeling more motivated, I upload a video to YouTube and then go back to post the link on an iNat obs. But, this is often more effort than I can get behind. Nobody else may care to see my YouTube videos, but at least it might give an IDer some helpful data. It just takes an unfortunate amount of time to do that whole process. Creating an animated GIF would be much faster. However, as @bouteloua points out, the iNat staff has been warning about disabling animated GIFs for years now (dunno why? What’s the harm?)

I feel like a video representation (gif or otherwise) adds so much more info than even 5-6 or more photos might do.

(Edited to correct some typos and appalling grammar)

3 Likes

Is it just animated GIFs that won’t be supported, or even non-animated GIFs?

i assume that if / whenever all GIFs get converted, the system will also stop accepting new GIFs as images.

1 Like

Thanks so much to everyone who replied! Thanks for the information and suggestions. I have some things to try now!

2 Likes

I didn’t know iNat is going to phase out gifs.

In that case, it would be great to see more support for having embedded videos as this feature request by @jamesray suggests :
https://forum.inaturalist.org/t/embed-media-by-pasting-a-url-to-the-media-e-g-from-youtube/8272

I’m wondering why it isn’t already possible to embed videos easily in markdown - it seems it is on other sites … is it a complex thing to add? or just contentious?

@bouteloua explained why embedding isn’t a good idea earlier in that thread https://forum.inaturalist.org/t/embed-media-by-pasting-a-url-to-the-media-e-g-from-youtube/8272/4:

I think as hard as it is, everything should be hosted on iNat itself. From before local copies of images were created, there are thousands of observations that used to be research grade but no longer are because the link with Facebook, Picasa, or Flickr is no longer working (i.e. user deleted the original image or revoked access).

That’s an opinion I share. Materials that are a part of the observation should be a part of the iNat data to keep it self-contained and internally consistent.

In the mobile app issue linked above, @kueda gave his opinion that supporting uploads of video file formats would be the way to go, but “it’s a heavy (and potential expensive) lift”.

Personally, I have adopted the workflow for my vids of:

  • load into OpenShot, strip audio, crop start & end, deshake, and zoom to crop the wobbly bits in the margins that result from deshaking
  • upload to my YouTube channel
  • add a link in the description when I submit the corresponding observation with one or more still cover photos (usually taken around the same time as the video - but if unavailable, I use one or more screenshots from the vid; usually one will suffice for ID purposes)
  • add a “Video Link” field to that observation that also contains the YouTube link
  • edit the YouTube video description to link back to the iNat observation

This may seem like a lot to go through vs. doing a gif, but honestly, I find making a good gif fiddly, and results in a lower quality rendering of what I’m trying to depict in the video. I agree with iNat’s reasons for not wanting to support animated gifs, and not wanting to support embeds either. Links to YouTube, for me, are the best compromise until such time as iNat can actually support video formats (though I have no illusions of that happening anytime soon).

In my workflow I mention including a still photo or photos to establish an ID. If it later comes to light that these don’t contain enough to establish the ID, I can always take more screenshots and add them.

Finally, if my YouTube account ever goes away, well, that would suck. With luck, I’d still be around to put the originals on some other platform and fix all of those broken links. If not … I did my best to make sure each observation didn’t depend on any linked video, but was merely enhanced by it.

7 Likes

I’m confused. If you anyway use a link to Youtube, whether it’s visually embedded or not, the argument from Bouteloua with regard to dead links becomes null and void.

If gifs are being phased out and video a long way off, the temporary solution of embedding videos makes a lot more sense than offsite links which are not embedded at least.

The only arguments I can see against embedding are with regard to

  • the time it takes to implement from the programming side
  • how it impacts the individual pages loading

I would like you to consider in addition to those arguments against embedding, that there’s a psychological impact resulting from embedding: it makes the video seem like a part of the observation potentially leading to faulty reasoning about what the iNat data contains. When you follow a link to another site, leaving iNaturalist, isn’t it clearer that it is not a part of the iNat data? I think that’s a distinction that’s not as clear with embedding. You might consider that evidence when making an ID, but as a conscientious iNat IDer, if you based your ID on what you could see in materials not directly on the iNat site, I think you would be more inclined to say so in the comments, and then I could fix it (by adding more screenshots).

In short, I’m depending on the community to spot data inconsistencies that result in my use of links to YouTube videos. I’m also doing my level best to ensure enough stills are included on the observation itself to establish an ID. It’s not perfect, but it’s the best we have so far. And no, embeds do not, in my opinion, make this any easier but rather the opposite, as it diminishes incentive to point out those inconsistencies and have them fixed.

2 Likes

By the way, if you want to check my video observations, see https://www.inaturalist.org/observations?place_id=any&user_id=benarmstrong&verifiable=any&field:Video%20link and see if any of those couldn’t be IDed on the still photos alone.

A handful of those will have animated gifs as well, but I assume that those will eventually be replaced with still images. My plan, if-and-when that happens, would be to supplement those observations with additional stills as needed.

1 Like

My workaround would be to extract several consecutive frames from the video then selecting a series of image frames that best reflects the movements that can help ID the ant. (NB: I don’t know anything about ant movement).

Many image/video software programs can extract image frames from a video file.
(I have in the past used Irfanview and ACDSee to do this but with avi files, not mp4.)

The following video on shows how to do this with VLC, which is free and open source : https://www.youtube.com/watch?v=NIzWZg02kHU

Finally, instead of video, if you can do this with your camera, try taking a series burst shots instead?

1 Like

This has been discussed before. Here is one gem of a quote from a previous thread on this very topic:

5 Likes

I’ve only recently have begun to capture most of my macro subjects (the quicker ones, at least) using short 4k video clips. I then ‘scrub’ through these in a video editor and pull out the best frames to post as stills in iNaturalist.

This experience has given me a bit of a better sense about how video CAN contribute to the identification process but it’s also shown me that like stills, it’s all in the quality of the media that’s posted. Poor lighting, cropping, motion blur – these (and more) are all problems shared with stills but video can have even more problems in some respects. On top of that, you generally need more (possibly more expensive) software and the skills to use it to get decently edited stuff. But there’s also some unique things to video that need to be considered:

How short should they be to be more useful to an Identifier than just posting the frames? How much time in total is either added or removed from the IDers ‘caseload’ if you make them watch anything over say, 10 seconds?

Another unexpected discovery for me was that when you extract a video frame as a still from a video file, normally no date/time or mostly any metafile gets stored with the frame image. I’m constantly having to refer to the stills I also have taken around a subject to get that into iNat.

But the difference is that for me at least, having the camera taking 30 frames a second while pursuing a swift-footed or flighty subject has given me much more success than I was geting previously even with mutli-frame shooting. Plus, there’s the added benefit of being able to use a short video scan of a ‘still’ or very slow resting subject, to capture enough frames to create a focus-stack image. That’s been a real great advantage.

But I look at the selection of the best frames from these clips as a form of curation, for the identifiers and try to post only the best – in terms of identification – choices to aid the identifiers. I don’t always succeed, but I am trying to be conscious of the demands on the time which they all generously offer.

Which is also why I would be a little nervous in supporting across the board video support and even links to third-party video hosts like YouTube. Do we really want to add even MORE steps and time to the Identifiers in-baskets? Don’t forget, many of those platforms will be pumping in ads, or auto-starting the videos when you’re not ready plus the controls and/or the resolutions are not so convenient as the customizable identification setup available with the iNat site.

All that kind of stuff will not make the task of identification any faster, I’m sure even when the videos ARE properly edited and prepped (no scrolling needed, cropped, tuned, etc). And for videos that are not edited at all? Even a bigger waste of time than poorly prepped stills.

Finally, let’s not forget that a huge chunk of energy is required to store, transfer, and view video files compared to still shots. Would it be right to add from an ecological point, the collosal power consumption that the web now commands by creating more animated waste pixels?

When you think of how much power now is commited into people sharing dog and cat videos (and other junk), you have to wonder if this is something that will help explain the planet’s situation to our descendants 100 years from now.

4 Likes

Well, that’s one person’s opinion. No disrespect, but it might be a minority opinion. I am old, but younger people have grown up in a video dense world and seem less likely to share that opinion. I do sympathize with data burden of videos, though.

2 Likes