Guess who thought, “I’ve heard all the stories about the complications of submitting subscriptions for review for the first time in an iOS app, so I’m sure I will have no issues.”

This guy.

Guess who was wrong.

This guy.

It’s me. I’m the problem it’s me.

(But it’s also App Store Connect)

Last weekend we went on a hike at Brown Mountain in Tucson Mountains. It was recommended as a sunset hike so we went later than usual.

The hike is a loop and has a quick ascent going clockwise, then some ups and downs with the last mile or so pretty much flat. It was a very enjoyable hike that had wonderful views, and provided a nice workout with the variations.

Activity Stats and Map

Distance: 4.4 miles
Elevation Gain: 753 ft
Time (including breaks): 2h 24m

I’ve been working on getting a subscription option set up in Mimi so folks can provide support for ongoing development with a sliding scale. Until that is live in the app store, I can’t launch the alt text generation feature, but it will remain available in the beta.

Gradually, then suddenly coming to the conclusion that the only way forward for Mimi Uploader is via a subscription model to support development and ongoing costs. And to offer that via various tiers to emulate a sliding scale for supporters.

It is tricky to make a subscription service or otherwise charge specifically for the AI image description feature even though it has ongoing cost. That’s because in many scenarios you don’t get the service you pay for, and it is out of my control. For example:

• This API is in preview. It may be turned off and never become a regular API.
• OpenAI currently rate limits the API use. Once several people are using this regularly, it is entirely possible to reach that limit.
• OpenAI’s service may be down.

So a subscription, if offered, is about supporting the app, which in turn supports this feature, and more in the future, in a best-effort way. It cannot be a guarantee of service.

How do you feel about that?

📷 Sunrise and Sunset

A new beta of Mimi Uploader is now out. It adds a new feature to generate a draft of an image description for use as alt text. An example is at the end of this post. This feature is opt-in. You have to ask for an image description to be generated each time. If you don’t then the app behaves as before.

Thank you for trying this out in the beta, which you can join at this link.

The way it works is that you tap the “Generate” button in the alt text editor. This sends a link to your previously uploaded photo to the service, which responds with the description. At the current time this feature only generates a description in English. Would you like to see an additional feature to translate the description into a different language? If so, please let me know the name of the language(s).

The API used for this feature is a paid service from a third-party. Each tap on “Generate” incurs a cost. If you would like to support this feature, in beta and afterwards, please give a donation directly, or drop a tip using the Settings page in the app. Many thanks to those of you who have already done so. All donations and tips are greatly appreciated.

As I prepare to launch this, I want to add that ever since I added the alt text feature, it’s sometimes been a barrier to me posting batches of photos because I wasn’t always ready to write all the descriptions. It felt like a chore sometimes. This feature greatly removes that friction. I see it as a positive step in my blogging (which often has batches of photos), as well as a more accessible web.

And now for an example illustrating this feature

Some weeks ago I shared the following photo:

The alt text I wrote at the time was:

Colorful morning sunrise in the background and silhouetted porch in the foreground. In between there is a silhouette of a Joshua Tree.

Using this feature, the alt text generated was:

A picturesque sunset with vibrant shades of pink, orange, and blue over a desert landscape with silhouettes of Joshua trees. Foreground includes the silhouette of a barbecue grill and patio with hanging light bulbs under a corrugated metal awning.

I’m very impressed by this description. It is spot on for everything except confusing a photo of a sunrise for a sunset, which is something many people do as well.

Additionally, every alt text in this post full of photos was generated using this feature.

I hope you find this feature useful in your blogging as well. I look forward to your feedback.

A new round of beta for Mimi Uploader is about to start. This one adds a feature to generate alt text for a photo based on OpenAI’s GPT 4 Vision model, which is a paid service. So any cash you can throw in the tip-jar in the app store version of the app would be appreciated.

A different sunrise view today 😍

📷 A bit of Fall at Mt. Tabor yesterday.