Image Generator is AIUNIFY PROS SUITE’s specialized AI Marketing workspace for creating visual marketing assets from written prompts and optional reusable Product and Avatar references.
The workspace itself is titled:
Text-to-Image Generator
with the description:
“Create stunning marketing visuals with AI.”
The current Image Generator supports subscriber-facing workflows for:
Navigate to:
Sidebar → AI Marketing → Image Generator
The current route is:
/ai-marketing/image-generator
The Tool also appears on the Marketing Studio Dashboard under:
Text-to-Image.
The system currently uses two related Names.
Image Generator
Text-to-Image
and:
Text-to-Image Generator
They refer to the same current workspace.
Image Generator checks whether the subscriber is authenticated.
If the session is unavailable, the page redirects toward:
/login.
The Image Generator page performs its own authentication check.
The supplied source does not show a separate page-level FeatureGuard wrapper around this specialized generator.
Normal Sidebar, Plan, or system configuration can still affect whether the Tool is available to the subscriber.
This Tool does not use the Standard AI Generator workspace from Chapter 12.
It uses a specialized Marketing media-generation service with:
The current Image Generator contains two primary columns.
Generation controls
Image Preview/result area
Below them is:
Generation History.
The current creation controls include:
The image Prompt section is labeled:
Imagine...
The current example is:
“E.g. A futuristic luxury watch displayed on a marble pedestal with dramatic lighting...”
Image Generator will not intentionally start a generation with an empty Prompt.
If you attempt to generate without one, the exact current message is:
Please enter a prompt.
The Textarea does not establish:
maxLengthKeep Prompts focused and descriptive.
Once generation is active, the Prompt Textarea is disabled until the operation finishes or fails.
This prevents the visible Prompt from being edited while the current task is processing.
Useful Prompt elements include:
For example:
“Business software ad.”
leaves most creative decisions to the generator.
For example:
“Premium enterprise software advertisement showing a diverse sales leadership team reviewing an AI-powered business dashboard in a modern executive office, dramatic cinematic lighting, sophisticated black and burgundy accents, polished commercial photography, clean negative space for headline placement.”
This gives the generator much stronger visual direction.
If a particular subject matters, identify:
Useful directions can include:
Examples:
If brand consistency matters, state the desired palette directly.
For example:
“Use burgundy, black, white, and subtle gold accents. Avoid bright blue as the dominant color.”
You can also explain the intended purpose.
For example:
This can help the generated composition match the intended business use.
If the generated image contains important:
inspect them carefully.
AI-generated visual text can contain spelling or character errors.
Below the Prompt, Image Generator includes:
Select Product Reference.
This allows a previously uploaded Product image to guide generation.
You can generate an Image using only the written Prompt.
A Product Reference is not required.
If the Product you need is not already available, select:
Upload Product.
The Product Picker accepts:
image/*
and explicitly checks that the selected file's MIME type begins with:
image/.
If the file is not recognized as an Image, the exact current notification is:
Please upload an image file.
The current shared Product Picker source does not contain a subscriber-side file-size validation check.
Therefore, this manual does not invent a Product Reference size limit.
A server/service may still impose limits that are outside this sanitized client source.
The Product Picker reads the selected Image as browser Base64 data before sending it into the Marketing upload workflow.
When uploaded, the Product Name is derived from the file Name without the extension.
For example:
AIUNIFY_CRM_Dashboard.png
produces a Product reference Name based on:
AIUNIFY_CRM_Dashboard
Instead of:
IMG_4289.png
use:
AIUNIFY_Pro_Suite_Interface.png
This makes the reusable Product reference easier to recognize later.
When the upload succeeds, the current message is:
Product uploaded successfully.
The Product list is then refreshed.
Possible current messages include:
Upload failed
or the specific returned Error.
If the request itself encounters an Error, the current message is:
An error occurred during upload.
If the reusable Product collection is empty, the Picker displays:
No products uploaded yet.
Choose a Product thumbnail.
The selected Product receives:
Product Picker retrieves existing references from:
list-products.
This means uploaded Products are intended to be reusable across supported AI Marketing generators.
When a Product is selected, Image Generator:
“Feature the product shown in the provided reference photo prominently.”
The selected Product URL is actually sent with the generation request.
It therefore acts as an AI reference input rather than simply an interface label.
The generator is instructed to feature the Product reference.
The source does not establish deterministic pixel-perfect Product reproduction.
Inspect:
before publication.
The current Product Picker exposes selectable Product thumbnails but does not render a dedicated:
Clear Product
button in the component.
If you need a clean no-reference generation, the current UI may require reloading/re-entering the workspace if simply selecting another Product does not meet your need.
Image Generator also includes:
Select AI Avatar.
You can generate with:
The Avatar Picker retrieves the current Avatar collection from:
list-avatars.
If no reusable Avatar exists, the exact message is:
“No avatars found. Create one in the Avatar Studio.”
Select:
New Avatar
to open:
/ai-marketing/avatar-studio.
The detailed Avatar-generation workflow is covered in Chapter 28.
Choose an Avatar thumbnail.
The selected Avatar receives visual highlighting and a Check indicator.
If an Avatar is selected, Image Generator adds the Avatar image to the reference list and appends:
“Include the person shown in the provided reference photo as the main character.”
If both are selected, another instruction is appended:
“The person should be naturally interacting with or showcasing the product.”
A useful marketing workflow is:
Select Product
Select AI Avatar
Prompt:
“Create a premium enterprise campaign image showing the spokesperson demonstrating the Product in a modern business environment.”
↓
Generate
This tells the service to combine both reusable references.
When combining references, inspect carefully for:
The Image Generator itself does not expose a general control labeled:
Its current reusable reference inputs are specifically:
If you need a new Product reference:
Upload Product → Select it → Generate
The generator then uses that stored Product Image in the request.
The current Image Generator does not establish:
There is no source-verified:
Remove Background
function in this Image Generator workspace.
The current control is labeled:
Resolution.
However, its available values are aspect/size ratios rather than pixel dimensions.
The visible options are:
The current default is:
1:1.
Use:
Square (1:1)
for compositions such as:
Use:
Landscape (3:2)
when you want a wider composition.
Examples:
Use:
Portrait (2:3)
when you want a taller image.
Examples:
The selector does not expose values such as:
The selected value is sent to the Marketing service as:
options.size.
Therefore:
Do not interpret the current Resolution selector as a precise pixel-resolution control.
The current Text-to-Image Generator does not offer:
16:9
in this selector.
The verified choices are only:
Likewise, there is no current:
9:16
Image Generator selection.
The interface includes a second selector labeled:
Style.
The current options are:
The current component initializes Style as:
realistic
which corresponds to:
Photorealistic.
Although the Style selector is visible and its state changes when you choose a Style, the current handleGenerate request does not include the style state.
The request sends:
but not the selected Style.
This is an important implementation distinction.
You can visibly choose:
but the current client does not transmit that choice to /api/marketing.
Therefore:
The Style selector is currently visible UI, but its effect on generation is not source-verified.
Until the Style selection is confirmed to be wired into generation, write important visual-style instructions directly in the Prompt.
For example:
“Photorealistic commercial photography with natural skin texture and realistic lighting.”
or:
“Premium 3D render with realistic materials, ray-traced studio lighting, and high-end product visualization.”
If a particular aesthetic matters:
Select the desired Style
and also:
State the Style in the Prompt.
This is the safest current user workflow.
The primary action button is:
Generate Masterpiece.
Select:
Generate Masterpiece
after completing the Prompt and any desired reference/settings choices.
Once active, the button changes to:
Generating...
You cannot start another generation with that same main button while the current generating state is active.
The Image Generator sends:
action: generate-imagesizeisEnhance: trueto the specialized Marketing API.
The current client sends:
isEnhance: true
for Image generation.
There is no user-facing toggle to disable that enhancement behavior.
The current UI does not provide:
Image Generator does not submit the request through the normal /api/ai/stream text-generation workflow.
It posts to:
/api/marketing
with:
action: generate-image.
The Image Generator component does not retrieve or submit:
selectedModelin its generation request.
Therefore:
Changing the normal PROS SUITE Header AI Model is not source-verified to change Text-to-Image generation.
The current Image Generator does not expose a user-facing choice such as:
The Image Generator request sends ordinary JSON headers to /api/marketing.
It does not append the subscriber's provider API-Key headers used by the shared text-generation workflows.
Therefore, Chapter 11's BYOK AI Model configuration is not source-verified as controlling this Image-generation operation.
While an Image is processing, the current interface states:
“Kie.ai is processing your high-resolution marketing image.”
This indicates the current specialized media-generation workflow is integrated with Kie.ai at the user-interface level.
The initial generation request does not immediately return the final Image.
Instead, a successful start returns a:
taskId
which the client stores for later status checks.
When the initial task begins successfully, the current notification is:
Generation started. Please wait...
While a Task ID exists and generation remains active, Image Generator checks the status approximately every:
5 seconds.
The current client keeps polling until the provider/service reports:
successfailedor the component is otherwise removed.
Status checking happens automatically.
You do not need to select:
Check Status
manually.
Unlike Game Developer, Image Generator does not currently expose:
Stop Generation
or:
Cancel Generation.
Once a task starts, the client waits for a success/failure state.
The polling logic does not define a fixed user-facing timeout such as:
before automatically canceling the task.
During generation, the Preview displays:
Creating your vision...
Below that is:
“Kie.ai is processing your high-resolution marketing image.”
The current generating screen includes a horizontal Progress animation.
Its animation is configured to fill from 0% to 100% over:
30 seconds.
The animation is simply configured with a 30-second visual duration.
It is not linked to:
Therefore:
Do not interpret the animated bar as an exact generation-progress measurement.
Because task completion is controlled by 5-second status polling, the actual Image can still be processing even if the visual progress animation has reached the end.
Wait for the success notification/result.
When generation begins, Image Generator sets the previous current result to empty before starting the new task.
Therefore, preserve an important Image through:
before intentionally starting another generation if needed.
When polling reports success, the current client:
The exact success notification is:
Image generated successfully!
The current client explicitly uses:
status.resultUrls[0].
Therefore, the current result area is designed around one returned Image.
The Image Generator does not expose a control for:
The current result workflow uses one primary returned URL.
There is no source-verified subscriber setting for:
Seed.
The current interface does not include a separate:
Negative Prompt
field.
Put avoidance instructions into the main Prompt where appropriate.
Example:
“No text, no watermark, no blue background.”
There is no user-facing:
control.
The Image Generator does not expose a:
Steps
control.
There is no source-verified:
Generate 4 Variations
button.
Regeneration starts another full generation request.
If task polling returns an Error containing wording related to:
the Image Generator treats it as a safety-filter condition.
The current message is:
“Your prompt was flagged by the AI safety filter. Try rephrasing — avoid mentioning real people, brands, or sensitive topics.”
A safety rejection means the generation service declined the Prompt.
Revise the Prompt rather than repeatedly submitting identical text.
The current Error guidance specifically mentions avoiding:
real people
when a Prompt is flagged.
If a request involving a named public/person subject fails, rework the creative direction appropriately.
The current safety message also mentions:
brands.
If a brand-heavy Prompt is rejected, try describing the desired Product or visual characteristics in a more general way when appropriate.
The same message references:
sensitive topics.
Keep marketing-generation Prompts appropriate for the service.
The polling logic also detects the phrase:
credits insufficient.
This is distinct from the PROS SUITE Monthly Tokens display.
The current client exposes the possibility of a media-provider credit Error while the general Feature catalogue separately lists a PROS SUITE Token reference for Marketing Image generation.
These are not automatically the same accounting system.
In a specific demo-domain condition, insufficient provider credits cause:
“In the demo, this feature is disabled”.
This is demo-specific source logic and should not be interpreted as a standard production subscriber message.
If task polling reports a normal failure, Image Generator displays the returned Error or:
“Generation failed. Please try a different prompt.”
If the Marketing service refuses to start the generation, the current UI displays the returned Error or a message based on:
Failed to start generation.
If the initial response cannot be parsed as expected, the client creates an Error in this pattern:
Unexpected response from server ([status]).
If the request itself throws an Error, the fallback is:
Image generation request failed.
After successful generation, the completed Image appears in the large right-side result area.
The current result uses:
object-contain
so the complete Image is intended to fit inside the Preview rather than being cropped to fill it.
Before generation, the Preview displays:
Your creation will appear here
with:
“Fill in the prompt and settings on the left to start generating.”
When a result exists, two controls appear near its upper-right area:
Select the Download icon to save the current generated Image.
The direct result Download requests a filename in this pattern:
ai-suite-gen-[timestamp].png.
The direct Download action names the current result with a:
.png
extension.
The current result Download immediately displays:
Download started.
The underlying helper attempts to retrieve the generated asset through AIUNIFY's Download proxy and save it as a browser Blob.
If that fails, it opens the asset URL in a new tab.
If clicking Download opens an Image in a new tab instead of saving the file directly, the shared Download fallback likely occurred.
Use your browser's normal Save function when appropriate.
Beside Download is a Refresh-style control.
Selecting it calls the same:
handleGenerate
workflow again.
The regeneration action uses the current:
It creates a new generation task.
Regenerate does not send the current generated Image back as an image-editing input.
It simply runs another Text-to-Image request based on the current generation inputs.
Because handleGenerate clears resultImage at the beginning of every run, selecting Regenerate removes the current Preview while the replacement Image is being generated.
Before selecting Regenerate:
This reduces the chance of losing easy access to an approved version.
The current Refresh control immediately starts another generation.
There is no source-verified confirmation dialog.
Because Regenerate launches another full media-generation task, treat it as a new generation operation rather than a free Preview refresh.
The current Image itself is clickable and attempts to open Asset Viewer.
When you click the current result, the component sets:
viewerIndex = 0
but the Asset Viewer receives:
recentImages
rather than a one-item array containing resultImage.
The current result click relies on the first item in Generation History being the desired current Image.
If the backend's asset ordering does not place the newest result first, Asset Viewer could open a different Image.
If clicking the current large result opens the wrong historical Image:
This uses its exact History index.
Below the main generator is:
Generation History.
When the authenticated Image Generator opens, it requests:
list-assets&type=image.
Unlike Marketing Studio's eight-item mixed Image/Video recent view, Image Generator requests only assets with:
type=image.
The Image Generator sets the returned Image array directly as:
recentImages.
It does not apply the central Marketing Studio Dashboard's eight-asset limit.
The client does not perform its own date Sort on the returned Image array.
The visible History order depends on the order returned by the Marketing service.
After polling reports success, Image Generator automatically runs:
fetchRecentImages().
The newly created asset can therefore appear in Generation History without requiring the subscriber to reload the page manually.
If no Image assets are returned, the current message is:
No history found.
Each historical Image is displayed using the shared:
MarketingAssetCard.
Each Image History Card can display:
The Card converts the Image asset Type into uppercase, typically:
IMAGE.
The stored Prompt appears beneath the Image.
If there is no Prompt, the fallback is:
No prompt provided.
History Cards use the stored created_at timestamp to display relative time.
Hovering over a History Card exposes:
Although MarketingAssetCard can support Delete when an onDelete callback is supplied, Image Generator does not pass one to its History Cards.
Therefore:
Image Generation History does not currently expose Delete.
There is no source-verified:
control.
There is no:
Search Images
field in Generation History.
Since the page already requests only Image assets, no separate user filter is exposed.
There is no subscriber-facing sort control.
Generation History does not establish:
Click the Image Card or its Eye control.
The page stores the selected History index and opens Asset Viewer.
The shared Asset Viewer opens the selected Image in a dark full-screen/lightbox-style environment.
Asset Viewer loads the Zoom plugin.
This allows closer inspection of generated Images.
Inspect:
before approving an Image.
The MarketingAssetCard Download action saves an Image using a Name in this pattern:
marketing-image-[first 8 characters of asset ID].png.
The two Download paths use different filename patterns.
ai-suite-gen-[timestamp].png
marketing-image-[asset-id].png
The History Card also provides a Share-style control that copies the asset URL.
The exact notification is:
Link copied to clipboard.
The Share-style action copies:
asset.url
not the Image Prompt.
There is no dedicated:
Copy Prompt
button on the History Card.
The system simply copies the currently stored asset URL.
It does not create:
For approved business creative, download the actual file.
A remote media URL should not automatically be treated as permanent business storage.
Generation History is retrieved from the Marketing service rather than being stored only in the current React page state.
This establishes persistent Image Asset history across page visits, subject to the service retaining the record.
The client does not display a separate:
Save Image
button after successful generation.
Instead, success triggers a History refresh, showing that generated media is handled through the Marketing Asset service.
Because persistence is tied to the generator/service workflow, there is no visible:
Save to History
action.
The generated media result is not managed like Website Builder or Game Developer project drafts.
The system operates around generated Marketing Assets.
Even with History, download approved production assets.
This gives you an independent copy for:
The general PROS SUITE Feature catalogue currently lists:
Marketing Image Gen — 50 Tokens.
The Image Generator page itself does not display:
50 Tokens
beside Generate Masterpiece.
It also does not contain a direct client-side PROS SUITE Token-deduction call.
Final accounting occurs outside this sanitized client bundle.
Therefore:
50 Tokens is a source-verified feature catalogue/default reference, not proof that every current production Image request is permanently fixed at exactly 50 Tokens.
The page explicitly recognizes a provider Error containing:
credits insufficient.
Accordingly, specialized media-provider availability can be a separate operational consideration from the account's displayed Monthly Tokens.
The specialized Image page does not display:
inside the generation Card.
Use the normal PROS SUITE Monthly Tokens display for account-level monitoring.
Unlike the shared text-generation hook, the Image Generator's successful polling path refreshes the Image History but does not call the shared refreshUser() account function in this client source.
Therefore, the Sidebar Token display is not source-verified to refresh immediately as a direct consequence of Image completion without some other account refresh/navigation event.
Because no selectedModel value is read or submitted, changing between normal PROS SUITE language Models should not be documented as the current method for choosing the Image model.
The current media request does not forward the provider API-Key headers described in Chapter 11.
The specialized Marketing generation service controls this workflow from the client perspective.
There is no user-facing:
field inside the current workspace.
Before generating, decide whether the Image is intended for:
Then write the Prompt accordingly.
Avoid combining too many unrelated scenes into a single Prompt.
A focused Prompt generally gives the generator a clearer visual target.
If the Product must dominate the composition, say:
“The Product is the clear hero and occupies the center foreground.”
If the Avatar is primary:
“The spokesperson is the main subject, with the Product clearly visible in their hands.”
Because the visible Style dropdown is not currently transmitted in the generation request, repeat the desired style in the Prompt.
Example:
“Cinematic commercial photography, realistic materials, dramatic lighting.”
Choose:
based on composition.
Do not assume the selector guarantees a particular pixel output.
Changing the ratio afterward does not crop or transform the already generated Image through a source-verified editing workflow.
Choose the desired format before selecting Generate.
When advertising an actual Product:
A useful source Product Image should ideally have:
Even when using a reference, compare the generated result with the actual Product.
Check:
For campaign continuity, use the same approved Avatar across related creative where appropriate.
Compare:
before publishing.
AI media often struggles when a person is:
Zoom in before approval.
Without a dedicated Negative Prompt field, use clear Prompt instructions such as:
“No text, no watermark, no extra fingers, no blue lighting.”
Results are still generative, so verify rather than assume compliance.
Avoid entering:
into creative generation Prompts.
A good workflow is:
Generate
→ Inspect
→ Download if approved
→ Then Regenerate if another version is needed
Instead of treating Regenerate as overwriting the creative concept, use Generation History to compare multiple outputs.
Once you choose the strongest Image, preserve it outside PROS SUITE using Download.
Current generated Names are system-oriented.
Rename important assets externally.
For example:
AIUNIFY_Enterprise_Campaign_Hero_01.png
For marketing production:
hero-v1.pnghero-v2.pnghero-approved.pngcan make campaign management easier.
Before publishing, review:
A Prompt passing the AI safety filter does not mean the Image has been:
Those reviews remain separate.
If the creative uses recognizable brands or other intellectual property, make sure your intended use is appropriate.
Check:
The current Image page itself redirects unauthenticated users to Login.
Check the Prompt.
If it is empty, the current system displays:
Please enter a prompt.
The Prompt is intentionally disabled while generation is active.
Wait until the current task succeeds or fails.
The current client does not send the Style state in the generation request.
Write the desired style explicitly in your Prompt.
Again, reinforce:
“3D render”
or the detailed desired aesthetic in the Prompt.
The dropdown alone is not currently source-verified to control the request.
The field labeled Resolution currently selects an output ratio/size:
It does not expose 4K or explicit pixel dimensions.
16:9 is not one of the current Image Generator size choices.
Use one of the verified available options.
The current Image Generator does not provide a generic reference upload field.
Use:
Upload Product
for Product references or:
Avatar Studio / Select AI Avatar
for reusable person references.
Product Reference accepts Images.
Convert or extract the appropriate Product Image before uploading.
A successful upload refreshes the Product list.
If it does not appear:
The Name is derived from the uploaded filename.
Rename the source Image before uploading another Product reference.
Reference-guided generation is still generative.
Try:
and regenerate.
Select:
New Avatar
and create one through Avatar Studio.
Reference consistency is not pixel-perfect.
Use an approved Avatar reference and simplify the scene where necessary.
Make the interaction explicit.
Example:
“The spokesperson holds the Product naturally with both hands, Product facing the camera, realistic hand anatomy.”
The visual Progress bar is not actual provider progress.
The client continues checking task status every five seconds.
Wait for:
rather than relying on the animation.
The 30-second progress animation is cosmetic.
Actual completion is determined by the provider task status.
Continue waiting for the result or an Error message.
The current Image Generator has no source-verified Cancel/Stop button.
The current interface tells you to rephrase and avoid problematic:
references where applicable.
Revise the Prompt rather than retrying it unchanged.
This message comes from the specialized media generation path.
It can reflect provider/service credit availability and should not automatically be interpreted as the same thing as your PROS SUITE Monthly Token balance.
Try:
This indicates the generator did not receive a response in the expected format.
Retry after confirming your session/network remains active.
That is current behavior.
Starting another generation clears the current result while the new task runs.
Use History or Download to preserve approved output first.
The current Preview's Viewer is wired to History index 0 rather than directly to the current result URL.
Open the specific Image from Generation History instead.
Image Generator does not pass a Delete action into its Generation History Cards.
No Delete control is currently available there.
No Rename function is established.
Rename downloaded files externally.
No Search control exists in current Generation History.
The Download helper falls back to opening the asset URL if direct proxy Download fails.
Save the file from the browser where appropriate.
Copy Link copies the asset URL as Text.
It does not copy the Image pixels to the clipboard.
The Image Generator client does not submit the shared selectedModel.
This media generator is separate from the standard language-Model workflow.
The current Image Generator does not send the subscriber's normal provider-key headers to the specialized Marketing API.
Those BYOK settings are not source-verified as controlling this Tool.
Before generating:
During processing:
After completion:
Before starting another version:
You should now be able to:
| InputCurrent Support | |
| Text Prompt | Yes |
| Product Reference | Yes |
| Upload Product Image | Yes |
| AI Avatar Reference | Yes |
| Product + Avatar Together | Yes |
| Generic Reference Image Upload | No dedicated control |
| PDF Input | No |
| DOCX Input | No |
| Negative Prompt Field | No |
| Seed | No |
| Guidance Scale | No |
| Number of Images | No |
| Inpainting Mask | No |
| Visible ChoiceSent Value | |
| Square | 1:1 |
| Landscape | 3:2 |
| Portrait | 2:3 |
The current client sends this value under the generation request's size option.
| Visible StyleUI AvailableSent in Current Generation Request | ||
| Photorealistic | Yes | No source-verified transmission |
| Digital Art | Yes | No source-verified transmission |
| Cinematic | Yes | No source-verified transmission |
| 3D Render | Yes | No source-verified transmission |
The Style value exists in client state but is absent from the current generation-request body.
| ActionSupported | |
| View Result | Yes |
| Download | Yes |
| Regenerate | Yes |
| Edit Image | No |
| Crop | No |
| Resize Existing Result | No |
| Remove Background | No |
| Inpaint | No |
| Save Button | No separate button |
| Publish | No |
| ActionSupported | |
| View | Yes |
| Zoom | Yes |
| Download | Yes |
| Copy URL | Yes |
| View Prompt | Yes |
| Delete | No current Image History control |
| Rename | No |
| Search | No |
| Sort | No user control |
| Tag | No |
| Folder | No |
The current subscriber-facing source verifies:
The current source does not establish:
Several Image Generator behaviors deserve particular attention.
The Style dropdown is visible, but the selected value is not included in the current Marketing API request.
Use the Prompt to express important style requirements.
The current choices are 1:1, 3:2, and 2:3 rather than pixel dimensions.
The Tool calls /api/marketing, not the shared text-generation AI stream.
The selected PROS SUITE language Model is not part of this client's request.
Normal subscriber provider-key headers are absent from the Image-generation client call.
A task is created and polled every five seconds.
Its 30-second animation does not represent actual task completion.
The subscriber waits for success/failure after the request begins.
After successful generation, the generator refreshes its Image-specific History.
Use the current Image Generator in this order:
IMAGE GENERATOR
↓
Write:
PROMPT
↓
Optional:
SELECT PRODUCT REFERENCE
↓
Optional:
SELECT AI AVATAR
↓
Choose:
SQUARE / LANDSCAPE / PORTRAIT
↓
Choose visible Style
and also:
WRITE THE STYLE INTO THE PROMPT
↓
GENERATE MASTERPIECE
↓
Wait:
CREATING YOUR VISION...
↓
Automatic status polling
↓
IMAGE GENERATED SUCCESSFULLY!
↓
Inspect Current Result
↓
Open/Zoom when needed
↓
DOWNLOAD
↓
Review:
GENERATION HISTORY
↓
Create another version with:
REGENERATE
↓
Compare variations
↓
Download the approved final Image.
For a Product campaign:
UPLOAD CLEAN PRODUCT REFERENCE
↓
Select Product
↓
Optional Avatar
↓
Prompt:
Campaign Goal + Subject + Environment + Product Placement + Brand Colors + Composition + Lighting + Style
↓
Select ratio
↓
Generate
↓
Inspect Product fidelity
↓
Download approved variation
↓
Regenerate additional variations
↓
Use Generation History to compare outputs
↓
Select final campaign creative.
AIUNIFY PROS SUITE Image Generator is the specialized AI Marketing Tool at:
/ai-marketing/image-generator
and is presented inside the workspace as:
Text-to-Image Generator
with:
“Create stunning marketing visuals with AI.”
Subscribers begin with the Imagine... Prompt field and can optionally select reusable:
Product References can be uploaded as Images and then reused through the Product Picker. The uploaded Product Name is derived from its filename. AI Avatars are retrieved from Avatar Studio and can likewise be reused.
When a Product is selected, the generator adds its Image URL to the generation references and instructs the AI to feature the Product prominently. When an Avatar is selected, it instructs the AI to include that person as the main character. When both are selected, it additionally asks for the person to interact naturally with or showcase the Product.
The current Resolution selector provides:
but does not expose actual pixel dimensions.
The current Style selector visually offers:
but the current generation request does not submit the selected style value. Therefore, important style requirements should be stated directly in the Prompt rather than relying on the dropdown alone.
Selecting Generate Masterpiece submits a specialized Marketing request using:
isEnhance: trueThe request does not include the normal shared PROS SUITE selected language Model or the subscriber's standard provider API-Key headers.
Generation is asynchronous. A successful start returns a Task ID, after which the client polls approximately every five seconds. While processing, the page says:
“Creating your vision...”
and:
“Kie.ai is processing your high-resolution marketing image.”
The visible Progress animation lasts 30 seconds, but it is not tied to provider task progress and should not be interpreted as an exact completion percentage.
Successful completion produces:
Image generated successfully!
The first returned result URL becomes the current Image and Generation History is automatically refreshed.
The current result provides:
The direct Download uses a filename such as:
ai-suite-gen-[timestamp].png
while historical Image Cards use names such as:
marketing-image-[asset-id].png.
Image Generator also provides persistent:
Generation History
retrieved from the Marketing asset service. Historical Images can be:
but the current Image History does not provide Delete, Rename, Search, folders, or tags.
The generic PROS SUITE feature catalogue currently references 50 Tokens for Marketing Image generation. However, the Image Generator itself does not display a fixed 50-Token charge, and the final media-service billing/accounting logic is outside the sanitized subscriber source. The page also recognizes a separate provider Error for credits insufficient, so provider availability and PROS SUITE Tokens should not automatically be treated as the same accounting mechanism.
The recommended operating principle is:
Write a Precise Prompt → Select Product and/or Avatar References When Needed → Choose the Correct Image Ratio → Put Critical Style Instructions Directly in the Prompt → Generate → Wait for True Completion Rather Than the Cosmetic Progress Bar → Inspect the Image at Full Size → Verify Product, Avatar, Text, and Branding → Download Approved Results → Use Generation History to Compare Variations → Regenerate Only When You Intend to Create Another Full Image.