AI Pro: Filters, possibilities, and tips
What Are AI Pro Features?
AI Pro features offer advanced customization and capabilities designed to elevate your event or photo experience. Our integration with Nano Banana, from Gemini AI, impulses the AI generations to a next level. Create and expand your events with this top-tier feature.
Individual, or group picture, this feature adjusts to a very wide range of possible set-ups and theme events.
These features are not included in standard subscriptions and require the purchase of AI credits to access. By purchasing AI credits, users gain access to:
Enhanced AI-powered customizations
Additional tools for intricate photo modifications
Some of our filters include realistic generations, cartoon, or different artistic styles. The customization is not closely limited to these filters, but you can also add your custom ones, generate backgrounds with lighting masks, use filters for clothes swapping or write your own composition with a detailed prompt or using up to 4 custom files to be used in your AI generation.
With our newest addition, you can turn an image into a short video with custom prompt, generated with AI.
How to Purchase AI Credits
AI credits are required exclusively for these advanced features and do not affect your ability to use basic, included tools. Follow these steps to purchase AI credits and unlock AI Pro features:
Log in or create a new account at Touchpix.com.
Navigate to the shop, and AI Credits purchasing section.
Select the credit package that suits your requirements.
Complete the transaction securely to activate AI Pro features.
Once purchased, these credits can be used immediately to access premium capabilities within the app.
AI Filters
AI Pro has many different filters, and each of them has its own use for different outputs. There are 7 different filters to create infinite combinations and results to make a unique event.
Built-in filters
Offering three different built-in filters, you can choose among +100 predefined filters. These scenes are already generated and tuned by us to offer a great filter without the need of changing prompts or using more credits to create a basic generation.
This basic filters have a great output, however they're meant to generate only one person per picture. You can choose among:
Realistic filters
Cartoon filters
Art filters
Custom filter
Build your custom filter using a detailed prompt and an image if desired to create a personalized generation. This option will create the AI-generated character(s) for the faceswap as well as the background, based on the description given.
Choose a random prompt or one of our prompt library if you need some inspiration!
Background
This filter does not change the person in frame, but rather replaces the background digitally. In this option, you can add a custom background and a lighting mask to mix them and create a unique effect.
Clothes
A dedicated filter for clothes swapping. It does not replace or change the generated person, but changes the clothes the person is wearing with no other changes. Ideal for fashion, sports, or corporate events.
Faceswap Image
An option that gives you the ability to swap the face of the person in frame with the image uploaded. Simply choose your image to which you will perform the face replacement and the AI will take care of the rest.
Custom composition
One of the most complete filters. Here you can use up to 4 custom images as references to your final output as well as a custom prompt, where you can detail how the composition will be as a final result.
The AI engine will use your description to mix up the 4 images and create the perfect scene.
Video generation
Generate a stunning video with only a prompt. This prompt should be as detailed as possible, as any missing details or despcriptions will result into failed frames or erratic generations.
Choose a video duration of 4, 6, or 8 seconds, your aspect ratio (16:9 or 9:16), and choose if an audio shall be generated as well.
Ultra mode
Ultra mode is an additional option that enables the best models for content creation. The AI will essentially work on the same way, however, the AI models that will be used to generate these results will be the best ones for image generation, and the results will be even better.
The cost in credit consumption adds by 1 on each filter, while video generation remains with the same cost.
Filters cost
For each different filter and type of generation, there are also different credit costs:
Built-in Filters: 1 credit per generation
Custom Filter: 1 credit per generation
Background: 2 credits per generation
Clothes: 1 credit per generation
Faceswap Image: 1 credit per generation
Custom Composition: 2 credits per generation
Video Generation: 10 credits per generation
*Ultra mode adds 1 credit to the cost of each filter except video generation, that remains on 10 credits.
Tips and recommendations
AI image generators translate text into visuals by matching your prompt against millions of learned concepts, patterns, and image relationships. Instead of "thinking" like a human artist, the model acts as an engine that fills in missing information using probability. When a prompt is short or vague, the AI is forced to guess critical variables—such as lighting, camera angle, facial structure, skin texture, and composition.
In complex tasks like a face swap, vague instructions like "swap face from Image 1 to Image 2", for instance, usually result in uncanny or broken images. The AI may awkwardly paste the source facial features onto the target body without adjusting for matching skin tones, shadow directions, or head angles. Without explicit directions on what to modify and what to lock down, the model often blends the two identities halfway, warps the background, or replaces the target clothing and posture.
Different AI image functions require tailored prompts because each task relies on a distinct spatial and logical mechanism. These are some tips:
Background Generation: Requires broad, environmental descriptors. The prompt should define depth of field ("shallow depth of field with soft bokeh"), atmospheric lighting ("dappled sunlight filtering through autumn leaves"), and spatial context so the model generates a background that naturally frames the subject without stealing focus or introducing harsh scale mismatches.
Custom Multi-Image Composition (up to 4 images): Relies on reference mapping and positional layout. Because processing up to four images (e.g., combining a person from Image 1, an outfit from Image 2, a product from Image 3, and a background from Image 4) forces the AI to merge multiple distinct attention layers, your prompt must act as a blueprint. You need to explicitly assign roles to each reference—for instance: "Place the subject from Image 1 wearing the jacket from Image 2 into the street setting from Image 3, holding the item from Image 4, matching the warm lighting and perspective across all elements".
Video generation: Prompting for video requires controlling both space and time: state the subject and its specific movement (e.g., "a woman walking briskly through a crowded rain-slicked street"), the camera direction and speed (e.g., "cinematic low-angle dolly shot tracking backward at medium speed"), and the cinematic atmospheric lighting and frame rate in a concise prompt to give the AI a clear trajectory and prevent unnatural subject morphing, temporal jitter, or broken physics.
Prompt length and core details to add
Optimal Length: 2 to 4 detailed sentences (around 30 to 70 words).
Reference Mapping: Explicitly state which photo serves as the identity source and which serves as the background/body reference (e.g., "Use Image 1 for facial identity, and Image 2 for the target scene").
Transfer & Preservation Scope: Specify what must change and what must stay locked. Tell the AI to keep the target image's hair outline, pose, clothing, background, and lighting, while applying only the facial identity of the source image.
Photographic Environment: Describe the lighting (e.g., "soft studio lighting" or "golden hour backlighting"), camera angle ("85mm eye-level portrait"), and skin details ("natural skin texture with seamless blending and accurate white balance") to prevent a plastic, over-smoothed look.
Style / theme: Define the desired aesthetic for the final image.
"Cyberpunk neon portrait," "Vintage 1950s comic book," "Oil painting on canvas," "Realistic 3D Disney Pixar character."
Specific edits: Describe the changes you want to apply to the scene, background, or outfit.
Technical details: Refine the artistic or photographic quality.
Source photo: Use a clear, well-lit selfie or portrait. Front-facing photos work best.
Enable "Ultra mode": Use our new enhanced mode to enable the best models for content creation with a higher cost in credit consumption.
Ultimately, prompt engineering is an iterative process where the user maintains complete creative control over the final outcome. Because AI models generate visuals based on probabilistic interpretations of your words, achieving your exact vision depends on how effectively you guide, refine, and adapt your instructions to your specific taste and needs. Always keep in mind that AI image and video generation is rarely a "one-click" solution; it often requires a small amount of fine-tuning—such as tweaking descriptors, adjusting camera angles, or running a few fast iterations—to train the output toward your desired result.


