You may review the items below for Frequently Asked Questions regarding AI:
- Can I try AI free of charge?
- Yes! If you have not added the AI add-on to your PBU license, you can still trial AI. In the AI tab, select 'auto-apply' or 'user-select' and select your filter(s). When you receive a prompt to add AI to your license, select 'No'. The AI will process but have watermarks on the photos. Note: You will still need to create a Photo Party Upload account (if you do not already have one).
- If you are recording AI Video you will be limited to 5 renders at 720P. Once you have rendered 5 AI videos you will be prompted to purchase a license with the AI Add-On. If you already have a PBU license, you will only need to purchase the AI Add-On.
- Once you have procured a PBU license with AI Add-On you can render videos between 720P and 4K, between 4-8 seconds long.
- Are there additional fees for AI processing?
- The only extra fees (once you have added the AI Add-On to your license) are the following processing fees:
- Stable Diffusion: $0.25 per 512x512 image and $0.50 per 720x720 image.
- Gemini Nano-Banana: $0.25 per image
- Gemini Nano-Banana 2: $0.25 per image
- Gemini Pro: $.50 per 1K and 2K image, $0.75 per 4K image
- AI Video (Gemini NB/NB2/Pro) - between $1-$3 per render, depending on resolution and length
- These charges will be added to your event and billed on the same cycle.
- The only extra fees (once you have added the AI Add-On to your license) are the following processing fees:
- How long do AI images take to render?
- Most images will process in under 30 seconds. This could take longer (several minutes) depending on a number of factors such as: the image resolution, your internet connection, the volume and frequency images are being captured, the current load on our servers and how many images are being processed per photo captured.
- Processing Speed follows (Fastest to Slowest): Nano-Banana -> Nano-Banana2 -> Gemini Pro
- Videos can take longer for the same reasons. A strong internet connection is very important to render images/video onsite, otherwise they will queue up and can be sent later
- View our AI rendering model differences HERE
- If your internet connection is strong enough, renders should appear almost instantly on the PBU sharing screen:
- Can I produce multiple AI renderings from one photo?
- Yes! You can choose to create up to 4 renderings per photo captured, which can be configured in the AI tab as shown below:
- Note: This can also be done in the PPU Kiosk App (Windows Version) by toggling Enable AI photo resending in the Preferences tab. This will automatically produce two AI renderings for each photo captured, allowing guests to choose between the two (or both)
- Yes! You can choose to create up to 4 renderings per photo captured, which can be configured in the AI tab as shown below:
- Can I share images via email/text before they finish processing?
- Yes! You can select 'Allow share while Processing' and guests will receive their AI image(s) once the rendering completes:
- If using the Gallery Bar or Kiosk App, Images will appear with a yellow 'Processing' badge which can be shared right away
- Yes! You can select 'Allow share while Processing' and guests will receive their AI image(s) once the rendering completes:
- Will PBU produce NSFW content?
- It is exceedingly rare for NSFW content to be created and we routinely update our filters to prevent this from happening with the end-goal being that it is impossible. Face Swap filters will not produce NSFW material, as it is only swapping the face of an existing image. Generative AI can produce a variety of content depending on the language used in your prompts. Adding verbiage such as 'Modest' to the end of your Positive Prompt and 'NSFW' to end of your Negative Prompt can also help mitigate any unwelcome renderings. If a rendered image appears all black, that is because NSFW content was created and the final rendering was blocked.
- Why are some of my AI images coming out all black?
- This happens when the NSFW filter detects inappropriate nudity. The image will be replaced with an all black image on the server before it is sent back. If this is happening regularly, update your prompt so it will not make the NSFW content in the first place. We recommend adding "modest" to the positive prompt and/or "nsfw" to the negative prompt.
- Can I apply branding or overlays to AI generated content?
- Yes! AI filters can now be applied in the Page Designer. The AI filter(s) can be applied in the Photo Holder(s), and overlays can be applied the same way as non-AI photos. See our guide HERE for details.
Here is an example using our 'Through The Years' filter:
- Can I create non-square AI renderings?
- Yes!
- When you create a Face Swap filter using Stable Diffusion, you can load a target image that is Landscape or Portrait instead of square. We recommend sticking with 1200x1800 images so they can print natively. See resulting example below:
- Yes!
- If using Gemini (NB1/NB2/Pro) the render will match the aspect ratio of:
- The dropdown option set in the Aspect Ratio field of the AI filter
- The photo holder if you are embedding the AI image into a Page Design layout
- Why are the Default AI prompts not appearing in the AI tab?
If our default AI prompts are not appearing in the AI tab, please try the following method:
In Preferences, select 'Restore Generic':
Then click 'OK':
- You should now have a mixture of our default Face-Swap and Generative AI filters in your AI Tab:
The latest available presets are: Ugly Christmas Sweater, Winter Wonderland, Underwater, Christmas Elf, Witch, Frankenstein's Monster, Zombie, 1950s Art, Cyberpunk, Barbie, Superhero, Younger, Older, Steampunk, Painting, Golf, Cartoon and Punk
- Can I create AI group pictures?
Yes! You can take group pictures and get awesome results with PBU AI. But note, group pictures can be more complex and variable than single or two person photos and you may need to do some additional work to make sure the outputs come out reliably.
When taking group pictures (with Generative filters) the results can vary since we are not able to adjust the variables for each person in the picture. See some examples below:
ORIGINAL
AI RESULTS
AI systems are more likely to cause errors or misgendering with group photos, but there are some things you can do to improve your outcomes. Some prompts will be fundamentally better for a wider range of people.
Examples:
- The "punk rock" prompt works really well for many combinations. Likely because there are a large variety of source images the ai was trained on for punk rock, men, women, couples, groups, etc. VS,
- The "Cowboy" prompt which does not work as well. We can still get some good results, but it is much more likely to turn a man & woman couple into both men, or a group into all men. Likely because the training data is much more weighted toward men (and that the word "cowboy" has "boy" in it, probably doesn't help).
Pro-tip: Use the variables feature to help define the guests. This is critical. For example, adding "Male & Female" and "Group" to options to cover all bases.
