[Feature Request] Images/Screenshots for each Application in the Description

[Satanarious]

It would be nice to have images/screenshots of each package/application on the description page.

I would also like to offer my help with collecting the images or links to the images. It would be a time-consuming task since the package list will be ever-growing. I think this would really help the UI and encourage the exploration of packages.

Here's a concept:

[marticliment]

Yes, it'd really improve, but the thing is, from where do we get them

[Satanarious]

Most Packages have GitHub links or Websites associated with them which have screenshots on them.For the Packages which don't, images/screenshots would be a google search away. I can create a seperate json file with package ids as key and lists of links to images as values.

[marticliment]

If you'd do that (really just the most common), it'd be nice, and screenshots will possibly make its way into WingetUI

[marticliment]

I guess the json schema could be somethign like

{
  "winget": {
    "packageId": ["UrlToScreenshot1", "urlToScreenshot2"],
    "packageId2": ["UrlToScreenshot3", "urlToScreenshot4"],
  },
  "scoop": {
    "packageI3": ["UrlToScreenshot5", "urlToScreenshot6"],
    "packageId4": ["UrlToScreenshot7", "urlToScreenshot8"],
  }

}

and could be perfectly stored in this same repo.

What do you think?

[Satanarious]

I was thinking of doing it exactly the same way.

[marticliment]

You can check out this file:
https://github.com/martinet101/WingetUI/blob/main/WebBasedData/screenshot-database.json

I'll add the feature when I can.

[gerardog]

Hi! I just saw this issue, and just let me express my two cents...

Manual collection of images is literally an infinite task. At some point humans will get bored and abandon the task.
You may automate the collection, but if you do so, why not instead think of on-the-fly mechanisms to fetch the images.

• For MsStore apps, you may scrap images from the store itself, using it's store id:
  https://apps.microsoft.com/store/detail/9NTXR16HNW1T

• If the project home page is rooted in GitHub, you can get the GitHub repository social media preview, by fetching the repo and filtering by <meta property="og:image" content="{image url}". This may work even in non-github pages.

Other cases, would need to think about, maybe fetching all images from project homepage and filtering those bigger than 200x200...

[Satanarious]

There are a lot of packages that I manually did today. I couldn't find screenshots/images for several packages even after using the google image search. Now, regarding automation:

• Mostly, all apps from scoop and hundreds of apps from Winget don't exist on the MsStore.
• We can't be sure if the repo owner would have uploaded a social media preview image.
• Besides all that, several apps don't even have homepages linked in the description and are marked as unknown.
• Also, I checked a few project homepages, and some have 1 or 2 while others had no screenshots at all. But I'll have to agree that many had screenshots for the app.

Just because this process has to be from the developer's end, the image attachment part. Since we are dealing with cli's here, that is not the case. So, it has to be either done manually or abandoned entirely.

I was thinking of doing them by setting a daily goal of a number of apps to be done. Once this is done, since Winget and scoop both will have packages being added to them constantly, then someone can create a script to check what apps have missing screenshots in the JSON file I proposed. This way, missing screenshots can be processed with each release.

[ppvnf]

Both manual and automatic ways of collecting images have advantages and disadvantages. As gerardog explained, there's no guarantee people will collect images for new packages and as Satanarious said, automation is likely to fail for several packages. So, why not have a manual collection of images with an automatic image fetcher as a fallback?

Edit: an automatic image fetcher would also make it easier to those manually searching images, as this would only be necessary when the automatic script failed.

[Satanarious]

How do you propose this image fetcher will differentiate between an actually valuable image (screenshot of the package) and random images (non-useful images) that show up when doing a general google image search