hermantenuki / ascii-generator.site Goto Github PK

It is ugly for now with those links and switchers on both sides, maybe drop down menu for links will be better.

For now, if image is deleted on server in "_images/temporary/" folder, but user still have it opened, settings will no more work and all will be frozen without any errors displaying. This looks not right.
(Images are auto-deleting in 24-25 hours after upload).

On resolution higher than full hd, footer has fixed position and it's bottom line is missing

https://sentry.io/share/issue/39099ccfa6ef408db78115c622f4cf31/
https://stackoverflow.com/a/49255449

Color output would be useful, especially if it enables copying shell formatting.

If user is on page with language code in url, like ascii-generator.site/ru/about/ - he can't change language manually with buttons in lower right corner.

P.S. Also when submitting ascii-generator.site/ru link with language code - it's working, but there is no / slash at the end.

{% url 'feedback_url' as URL_feedback_url %}
...
<a href="{{ URL_feedback_url }}">Feedback</a>

Leads to "About" page.

Maybe it's because first line out of {% block content %}, when last ins't.

Additional "Twitch mode" checkbox under image and text generators, so output arts can perfectly fill twitch chat (for the pastas).

In #39 numba was added to img2ascii generators.

On 4 cores system it is improving speed by 100%-300%, while on single-core system it actually works with 70%-100% speed.

Numba is disabled by default in project/settings.py, can be enabled with environment variable NUMBA = "True".

Need to improve this situation and make a project faster with numba on different configs without getting a lot of unstable problems and warnings.

Image to ASCII generator is working with caching to optimize "share" functionality and overall generator's usage.

On a server, for some unknown reason, it's working randomly and it can sometimes don't write in cache and can sometimes don't read from cache.

Locally it works fine.

Possible reasons:

• Default locmem cache - maybe should try memcached;
• Low amount of memory on a server - 512mb.

requirements.txt needs to contain only minimal-needed modules to run website locally, the other file requirements_dev.txt needs to contain all modules.

Image to ASCII generator is working with caching to optimize "share" functionality and overall generator's usage.

On a server, for some unknown reason, it's working randomly and it can sometimes don't write in cache and can sometimes don't read from cache.

Locally it works fine.

Possible reasons:

• Default locmem cache - maybe should try memcached;
• Low amount of memory on a server - 512mb.

For some reason, chromium rendering Unicode-blocks poorly regardless of what font is used.

If we try to create art with Unicode-blocks in a monospace font, chromium will always leave some weird lines between symbols.

For example, let's compare Google Chrome and Firefox browsers:

Google Chrome

Firefox Browser

I improved this issue by detecting Chromium in JavaScript and applying some styling:

if (window.chrome) {
    art_div.css({
        'letter-spacing': '-0.1em',
        'line-height': '1.2em',
        'transform': 'scale(1.2, 1) translateX(8%)'
    });
}

Let's look again at the examples:

Google Chrome (with styling fixes)

Firefox Browser again

Now you can see that situation is a lot better, but Chromium still displaying art differently. If you look at the letters in both arts (especially "b"), Chromium's blocks are blurring or, like, floating to the upper side.

For now, when arts is being saved, all the generated ASCII results is being saved separately in different row in database. So, for the Text to ASCII generator we need to save 300+ different rows separately, so as when we need to load them - we need to request 300+ different rows.

That's not optimized at all, need to migrate to some new DB structure, where all arts is saved in one row in some array or JSON form.

Due to LESS being used, there is no reason to keep not-minified compiled CSS code

Steps:

1. Upload some picture.png to "Image to ASCII";
2. Press "New Image" button;
3. Try to upload same picture.png.

In these steps, nothing will happen at the end. Image won't be uploaded.
If you will try to upload other_picture.png at the step 3, it will work normally.

Need to fix it by re-uploading image or just opening last already rendered result (maybe it would be better).

Small issue, just reminding myself to fix later.

All the enumerated action buttons could be pre-translated, instead of translating them again

Hover is having more priority than disabled

Images like image.JPG or image.PNG are being detected as wrong file types.

I have loved using the generator but it seems like for the last two days the site is down?

Minor suggestion, but would it be easy to fix the "pixel size" on the image2ascii page so that they are not in a floating position, and do not move as the pixel size changes?