increasing number of spelling errors in graphics
Just generated eight Vertical Journey Visuals (two sets of four) and I think 7 of the 8 have multiple spelling errors. Some of these could be possible to fix in Photoshop but takes both time and having to have Photoshop
Log in to comment and vote
Comments5
Sebastian Mourra
Aug 22, 2025
@kbuhlir thanks for this. What about the quality and aspect ratio settings outside the prompt? Could you confirm that for me? Also if you have examples of the images misspelled, that would be super helpful. You could send them to me in the support chat inside of Alphana to keep the images and custom instructions private. When you submit it, hit the “talk to a human” button and I will see it.
kbuhlir
Aug 22, 2025
Done. Love the product overall. The onboarding, continued usage examples, and your willingness to improve it helps immensely.
kbuhlir
Sep 19, 2025
@Sebastian Mourra I just sent another example of this happening in the same chat for you to see. Can you pick up the chat when you have a chance and take a look at the custom instructions as well?
Sebastian Mourra
Aug 22, 2025
Hi @kbuhlir sorry to hear this. Can you confirm the settings used in the AI Card when generating these? Can you take a screenshot when you click edit and send that here? Also, if you can provide the prompt and the images, I can take a look to see if I can help.
One important note with AI images is that the quality needs to be set to HIGH as the other options are more likely to result in misspelled images.
kbuhlir
Aug 22, 2025
@Sebastian Mourra, it was the default instructions for “Journey Visual: Vertical”. That’s a little more than I can fit in a screenshot but here is a copy/paste
===
Create a bold, visually rich infographic-style visual that presents a journey of lessons or mindset shifts from the video transcript or text context. Be specific in the insights pulled from the source content, not generic advice.
For the orchestrator/final prompt output for each image: Ensure the following is included in the final prompt: • Use a tall, vertical layout designed for scrolling — ideal for infographic-style content or carousel posts. • Image must be a bold, visually rich journey-style visual that shows a sequence of 5 to 7 insights or mindset shifts from the video transcript or text context. • Ensure all prompts for final image start with “Create a bold, visually rich journey-style visual…” • Represent the journey as a winding path, climbing trail, or scenic route with distinct stops or milestones. • Incorporate metaphorical visuals at each stop (e.g., signs for guidance, bridges for risk, mountains for growth, fog for uncertainty, etc.). • Vary the visuals and layout between each insight so that each section feels distinct but cohesive. • Include a short label or caption near each stop to summarize the insight or mental shift. • Style should feel immersive, premium, and designed for visual storytelling — light 3D effects, soft gradients, depth, and icons are encouraged, but should not overwhelm the message.
===
I pretty consistently get:
misspellings
duplicate words/phrases
in this one.
I’m also noticing on the default “Wall of Insights Visual” that I’m getting duplicate post-it notes.
I do have global instructions set but I do not see anything in those that would be causing this. Happy to share those with you privately if helpful. Although, AI does like to cause these issues when creating graphics, so it may just be a state of technology.