Like how you can copy-paste a YouTube video URL into a Facebook post and it will automatically embed / show the video, so too would it be handy to be able to do the same in the notes and comments of iNat observations.
Similarly, the same could apply for photos and audio. For instance, if I upload audio on xeno-canto, I could get the URL to the audio file, paste it in the observation, and it would be embedded / playable without having to go the URL.
A key advantage of doing this is that it makes it easier to share data between biota citizen science websites and elsewhere, without requiring to store data on the servers of both websites.
Related features that would add value to this feature, or can be considered an extension to it, would be to then include the embedded content of observations (including videos) in search results.
I second this. I put my videos on youtube and put a link to it in an obs description. Allowing ppl to view it from within the obs would be convenient. I would prefer the viewing window to be collapsed so as not to take up extra room.
Though I think this would be cool, I worry that it might decrease incentive to provide animated observations as evidence on the observation. Sometimes an animation is exactly what you need to illustrate a behaviour or clinch an ID, but unless it is provided as media on the iNat platform, does it count as āevidenceā? An embed for a link media on another platform would blur the lines and maybe discourage people from at least providing still images from their video along with the video link (thus disqualifying it from ever reaching RG).
For current support iNat has for uploading short/small video clips to observations, see: https://www.inaturalist.org/projects/animated-observations . The limitations of such media are sufficiently strict & the bar set high for people to create decent animated observations, though, that few people actually add them (0.00004%, or barely over 1K observations on iNat are in this project).
Related to that, I see no obvious / reliable way to find observations with videos linked in the description to find out exactly how many observations would benefit from this feature. The best I could come up with is a tags/description search for āvideoā, matching about 10K observations. That wonāt, however, find any that just had a link to youtube and no mention of the word āvideoā.
There are, I suppose, searches for field:Video%20Link, and a handful of similar fields, but it appears even fewer people (roughly 0.5K observations) are using those than people submitting animated media to the Animated Observations project.
So as much as I like the embed idea, I think improved support for uploading/searching for video observations is more beneficial to improving iNatās data & IDās supported by videos.
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 will be ideal, but until then, embedded youtube/ vimeo would be good. Iām one of the few ppl who (fairly) regularly make animated GIFs, but they have to be short and low quality to keep under the size limit. So even after creating a GIF, I still direct people to a youtube video to see the full encounter in video& sound.
I think the point made here, and one I agree with, is that itās not good for the iNat data to allow third parties to destroy data, which is effectively what allowing embeds would do. A terms of service change, or cancellation of accounts would trash all of those embeds.
If this feature request were implemented, some people would do the easy thing (link, and get an embed), but not have any actual media, because they think the link is the media. Right now, thereās no such confusion about the link being the media since itās only a link.
So while I do think the idea is cool, without some refinement (automatically keeping a low-quality transcode of the whole video in iNat & link back to the original? I donāt think iNat has the resources to do that, though) I donāt think it could work.
I donāt know what you mean by destroying data. Are you concerned that instead of putting up photos and GIFs ppl will start only putting up embedded videos? Well that would be a problem, but I donāt think it is being suggested that an obs will only have an embedded video.
You could also try to automatically get thumbnails from the video as images for the observation, or automatically try to download audio from the video (there are websites that provide this function, some of which I have successfully used before. Search results, Google) and include that in the observation, in addition to the option you mention of automatically including a low-quality copy or extract of the video.
Yes, iNat should in encourage users in the create observation template that if embedding videos, they should also include screenshots, if possible of key points in the video where the specimen is clearly shown, or audio, or just a screenshot/photo of where it was seen. You can only go so far at the moment with automation, AI, neural networks, etc. I donāt believe it is currently possible for a neural network to get screenshots/still captures of key points in a video. However, it seems like YouTube thumbnails can often include good screenshots, but this is certainly not guaranteed.
I mean even if the user added other media, if the YouTube link stops working, any prior ID based on examining that video would be based on evidence that no longer exists. Hence, data would have been destroyed in that case.
I am describing extended features that may accompany this feature, and can be posted in a separate thread, and are not necessary for this feature in order to improve the status quo.
If an observation has a dead video URL (or the video has been removed ), and no photos, screenshots of the video or audio is uploaded, and the URL is not backed up by iNat or made available otherwise (e.g. via a Wayback Archive link or a link to it stored and publicly available elsewhere) then the community can flag the observation as being no longer valid.
If the video is not backed up or screenshots arenāt provided by the author, then not only the author, but others users (particularly if the video has a permissive license,* although the default for YouTube is unfortunately all rights reserved), could also take screenshots of the video, save them elsewhere thatās publicly shareable (e.g. IPFS, Imgur, cloud storage, etc.) and post the link. * Note that even if the video has an all rights reserved copyright license, I canāt imagine it would be illegal to take screenshots from a publicly listed video, share them publicly, and link back to the source. For this scenario where there is a video, it may be helpful to also allow other users to upload photos and audio that are extracted from that observation. Moderators and other users can flag content that does not appear to be from the video, to prevent abuse of this feature.
The same feature would be useful for audio observations, for example where original audio content is uploaded, but the author hasnāt edited it, and the edited version may (or may not) help with trimmed silence, adjusting the gain, and filtering. Same thing for photos, e.g. where a photo may show an ant or something small or hard to see, and the image could be cropped or the specimen could be circled to see it clearly. While it may be preferable for authors to edit content before uploading, that may not happen, so providing the ability for other users to be the editors and then other users to moderate can help.
My understanding is that @jamesray is basically proposing that links to media be playable on the observation detail page, not that these video/audio embeds would count as the āofficialā iNat evidence of an observation. So for this observation of mine, which has a photo on iNat but a link to a Vimeo video in the description, the link would turn into an embedded video in the description? Is that correct?
I think Iād be OK with embedded video and audio in descriptions and comments as a way to make those more easily accessible on the page (how many people actually click on a video link in an observation?), but I donāt think they should count as the āofficialā evidence of the observation. Also, in my opinion, pages with embedded videos and audio files are not aesthetically pleasing. They also take longer to load, in my experience, and load times for observations are already kinda slow.
As someone who voted for this feature, I canāt speak for Jamesray, but what I support is not for embedded externally hosted videos/sounds/images to become part of an observations ācoreā (for want of a better word) media, but as you say tiwane, for us to be able to embed the videos in the comments/description in the same way as we can currently embed externally hosted and internal iNat images using the āimg src = WEBSITEā code. Re it slowing the loading of a page, I agree that could be an issue so it would be better if the video did not autoplay and perhaps the window should be collapsed by default.
The inclusion or continued exclusion of videos and externally sourced media can be considered orthogonal to the original issue. Including media, while backing it up somehow for continued accessibility, or providing the ability to flag inaccessible external content, can be considered as separate features to this particular request. Since evidence on iNat is used for the purpose of ID only, rather than analysing behavior, limited media is necessary, and so backing up external media like videos may be considered as ātoo much baggageā, unless a snippet was extracted and had the quality reduced.
If I read you correctly, you see videos as depictions of behavior and not for establishing an ID? Youāre probably right for most cases, but I feel there are a not insignificant number of cases where an ID could be established from the video alone.
For example, for a turkey vulture in flight I uploaded, in any one frame of the video, the species was unclear, but watching the video clip, it is recognizable immediately. Therefore, I went through the somewhat fiddly process of making an animated gif for it, then linked to the full quality video on youtube. The gif itself, I feel, is sufficient as evidence, but it seems from the low number of animated gifs uploaded to iNat, most people are not going to the bother of doing that.
My concerns stemmed from this admittedly less common case where externally linked media is furnished as evidence and yet probably shouldnāt be as it could be subsequently destroyed. Yes, that could be flagged as no longer valid, and if it only happened to one or two observations that might not be so bad, but a sudden change of terms of service or ban of one prolific uploader could make this a huge headache.
I do get there being separate issues, here. However, they should be considered together because they inter-relate: Though supporting embeds āfor secondary non-ID purposesā (like analysis of behaviour) might be done with this particular intent in mind, I doubt if users would understand that they should base their IDās solely on the non-externally-linked media. Human nature would kick in and those that saw the video would very likely be influenced in making their IDs based on what they saw (i.e. they could not āunseeā the video and consider only the data that iNat itself has).
I fully agree with you on that you can use a video for ID, and that a video can be more useful for ID than photos and/or audio, while your example is a good case in point.
Since you mention animated GIFs again, let me see if I can make a GIF from a video on my phone and upload it to an existing observation, then post back my findings. As mentioned, I prefer to do uploads on my phone. I installed a GIF Maker-Editor app, created a GIF from a video, and tried to upload it to an observation, but got the error: "Error importing photo(s): Can only import photo files with JPG/JPEG/PNG extensions. I have just created a separate feature request for that, which is here: https://forum.inaturalist.org/t/upload-gifs-from-the-android-app/8444/2.
I would like to add that:
the ability to embed a video is not that useful in itself, as it is not that hard to click on a link
adding support for including animated GIFs, videos, etc. in observations (as evidence) does seem useful
While I agree that if a Youtube user (or user of whatever the site that the media is hosted on) is banned, or there is a change in the terms of service of that site, this would be a pain. Nevertheless, I still think being able to include videos as evidence would be an improvement on the status quo, and backing up the video on Wayback Machine, iNat servers, or other solutions can help to remediate this problem.
I also agree that users can tend to base IDs not only on the evidence shown on iNat. I was trying to be diplomatic and agreeable. Yes, these issues should be considered together, holistic thinking is better than tunnel vision. Here is one example of that: https://www.inaturalist.org/observations/35000448.
You can try Movavi video editor - URL - been using it for quite a long time. Itās a great option if youāre looking for an app with simple tools and that is easy to understand.