How to Try Kindroid Vignette Without Changing Your Default

1. Record the default without changing it

Open the intended Kindroid's settings and inspect Models > Selfie Model. Write down its current value, then leave it alone. Expected result: a baseline you can compare later.

Privately note the selected Kin, date, default model and visible standard and paid selfie balances separately. Do not publish a settings screenshot. If the original value is unclear, stop before generating; a remembered preference is not a reliable baseline.

2. Enter photo generation and select the one-image override

The general selfie guide says to open the gallery using the image button at the chat's top, then choose Generate Photo Selfie.

The Vignette guide places the per-image choice under Additional controls > Selfie model. Select Vignette there. Expected result: Vignette selected for this request, while the default control remains untouched.

Do not substitute a video-model selector or switch Avatar slot. The general guide says switching slots changes the active avatar everywhere. Keep the existing avatar and its description unchanged. If your interface does not expose the documented choice, stop and consult official help rather than guessing another settings route.

3. Write one neutral, inspectable prompt

Use a short fictional scene with one action and one setting. Expected result: a request simple enough to assess without editing shared character instructions.

Suggested prompt: An original fictional adult aged 30, seated beside a plain window, holding a closed green notebook, wearing a gray sweater, soft daylight. This is an untested editorial example, not a quality promise.

Vignette prioritizes the prompt's beginning. Its body tags belong in Avatar Description, not the selfie prompt; face descriptors can work in either. Material after /// in Avatar Description is excluded from selfies. Do not move shared description text or add body tags there just for this trial. Leave elaborate tag experiments for a separate, deliberate editing session.

4. Check the cost, then submit at most one request

Read the actual allowance, charge or credit information available before submitting. Expected result: an accepted cost within your ceiling, or a decision not to generate.

The purchase guide distinguishes standard from paid selfie credits and says standard credits are consumed first for photo selfies. Exact purchase prices depend on country and platform at checkout. This does not establish your Vignette entitlement, a free allowance or a universal cost per image; video rates are not photo rates.

Prefer existing credits only when their use is clear and acceptable. Do not buy a pack or upgrade merely to finish the checklist. Stop at unclear deductions or an unexpected payment prompt. If you proceed, submit once; a slow request is not permission to queue duplicates.

5. Inspect the result without declaring a winner

Compare the image with your written request: adult appearance, seated action, notebook, sweater and window. Expected result: a brief record of matches, misses and uncertainty.

Mark missing or ambiguous details honestly. A pleasing image does not prove which model ran, unchanged settings or repeatable quality. If the output appears underage or otherwise inappropriate, stop; do not share it or pursue that direction.

If generation fails or stays unresolved, record that status and inspect the visible balances without resubmitting. Do not assume failed requests are uncharged or refunded. A later correction should be a separate conscious decision within your original ceiling, not an open-ended effort to recover spent credits.

6. Reopen the default and compare the exact value

Return to the same Kindroid's Models > Selfie Model setting and compare it with your baseline. Expected result: the identical original value, checked separately from the image.

Check even if generation failed; do not generate again to infer the default. If the value differs, record the discrepancy. Restore the recorded value only if intended and the control is clear, then reopen it to verify. Restoration is recovery, not evidence that nothing changed.

7. Stop when the evidence is incomplete

Finish after the image assessment and default check. Stop earlier for missing controls, unclear eligibility, unexpected costs or uncertain settings. Expected result: a bounded outcome, including unresolved checks.

For official support, describe your platform, time, selected model and observed symptom with minimal redacted evidence. Do not send private chats, passwords or billing details in an initial report. Do not delete the Kin, reset shared instructions or buy credits as an improvised fix.

Conclusion and documentation limits

Use the per-image selector, keep shared avatar settings fixed and verify the default directly afterward. Image appearance and default preservation are separate checks.

The four exact official sources were independently read through the web reader on October 28, 2026. Supplied direct HTTP captures returned a loading shell, not full documentation. This is documentation-only guidance: no signed-in access, payment, generation or successful default-preservation test was performed. Expected results above are checkpoints, not observed outcomes.

Frequently asked questions