A talking photo can make a real face deliver words that person never recorded. That power changes the approval question. A client sending an image is not enough. Before production, verify who controls the file, who is depicted, what the face will say, and where the result will appear.
Open the Pippit AI talking photo tool after both records are ready. This is a production control, not legal advice. Escalate disputed, political, deceptive, or unfamiliar uses.
Start With Two Proof Lanes
Do not place every permission question under one checkbox. The source image and the person in it create two separate proof lanes.
The source lane answers who made the photograph, who owns or licenses it, which file may be edited, and whether commercial animation is allowed. A client may own a copy without owning the copyright. A stock license may permit advertising but restrict synthetic edits. A photographer may approve an image while the subject has not approved the new performance.
The subject lane answers who the person is, who can authorize the portrayal, what script and voice are approved, and which contexts are allowed. A subject release for a still campaign does not automatically cover animation, translated speech, an endorsement, or future reuse.
Create one row for each lane. Add the evidence file, signer, date, scope, expiry, and contact owner. If either row is blank, the project is not ready. This simple split prevents a confident client email from being mistaken for complete permission.
Sort the Request Into One of Four Cases
The same consent process does not fit every source. Sort the request before discussing creative choices.
The client made the portrait
Ask who took it, whether another party holds rights, and whether the person agreed to synthetic speech. Request the original file and actual approval. Record whether the client may authorize this campaign or must return to the subject.
The client licensed the portrait
Read the license for media, territory, term, commercial use, alteration, and artificial intelligence restrictions. Confirm whether a model release exists and whether it matches the planned portrayal.
The subject is an employee, customer, or creator
Obtain specific approval from the person or authorized representative. Show the final words, voice, channel, term, language, and disclosure plan. Employment or an earlier collaboration is not permanent permission.
The subject is a public figure
Stop the normal production path. A public photo is not a blanket license. Identify whether the use is news, commentary, parody, education, advertising, advocacy, fundraising, or impersonation. Send the source, script, voice plan, placement, and disclosure decision to the responsible legal and editorial owners.
Do not solve uncertainty by making the face only slightly different. If viewers could reasonably identify the person, treat the identity question as real.
Test the Portrayal Before You Test the Animation
Consent depends on what the output communicates. Review the portrayal on paper before a talking photo makes it persuasive.
Read the script in the first person. Highlight claims about experience, health, money, performance, politics, beliefs, or product use. Ask whether the real person approved those words and could honestly stand behind them. A neutral headshot paired with "I recommend this treatment" still creates an endorsement.
Write down the intended voice source. A generic synthetic voice, an authorized recording, and a cloned voice create different expectations. Approval for the face does not automatically approve a voice that resembles the subject. Approval to translate the words does not automatically approve a new accent or emotional delivery.
Place the output in context. Record the account, caption, call to action, sponsorship label, and destination. A clip may become misleading beside a purchase button, donation request, political message, or fabricated quote.
Use a portrayal statement that a reviewer can approve in one sentence. For example: "This fifteen second video uses the approved headshot to deliver the attached product setup instructions in a neutral synthetic voice on the client help center for six months." If the team cannot write that sentence accurately, the scope is not settled.
Give the Client a Decision Packet, Not a Vague Warning
A useful client review shows the decision that must be made. Build a packet with five visible parts.
- 1
- Source preview. Show the exact portrait, its file name, source, creator, and license record. 2
- Portrayal preview. Include the locked script, voice description, language, facial framing, and planned duration. 3
- Placement preview. Show the channel, account, caption, surrounding message, call to action, territory, and run dates. 4
- Disclosure preview. State what label or explanation will appear, where it will appear, and who approved that choice. 5
- Decision page. Separate approval, requested change, and rejection. Name the signer and keep the date and version.
Do not ask, "Are you okay with AI?" That question is too broad to protect the subject or guide production. Ask the client to approve one defined use. If the script, voice, placement, or term changes, issue a revised decision page.
Use Hold Codes That Stop Ambiguous Work
Creative teams need a way to pause a project without arguing about legal conclusions. Add short hold codes to the production board.
Use S1 when source image authority is missing. Use S2 when subject permission is missing or too broad to verify. Use P1 when the script or implied endorsement changed. Use V1 when the voice source or voice approval is unclear. Use C1 when the campaign context, disclosure, channel, or term changed. Use E1 when a public figure or other elevated risk requires specialist review.
A hold code identifies the evidence needed to continue. The owner closes it by attaching the record and naming the reviewer. A deadline or convincing draft is not evidence.
This system also helps a client see why one talking photo can proceed while another cannot. The answer is tied to evidence, not to the confidence of the person requesting it.
Build and Release the Approved Version in Pippit
Once both proof lanes and the portrayal statement are complete, move the locked materials into Pippit.
- 1
- Upload the approved source. In Video generator, choose AI talking photo and add an image that meets the current minimum size. Confirm that the team has the right to use that exact file. 2
- Add only the approved performance. Enter the locked script or upload the approved audio. Select the intended voice and language. If pronunciation changes meaning, revise and return the new version for approval. 3
- Review the actual output. Check identity, words, voice, captions, facial motion, disclosure, crop, and surrounding campaign context. Send the export candidate, not a description of it, to the named approvers. 4
- Export one released version. Save the output file name, export date, approval record, and approved destinations together. A later translation, script edit, new crop, or new account is a new release decision.
For motion specific review after consent is clear, use the Pippit lip sync animation guide to examine timing without reopening identity questions. If the project needs a reusable presenter rather than one authorized still, compare the scope with the Pippit AI avatar generator before production.
Save Proof That Matches the Published File
An approval email becomes hard to use when the team cannot tell which export it covered. Create a release folder named for the campaign and final file. Store the source record, subject record, portrayal statement, script, audio, disclosure decision, final preview, approver names, approval dates, term, and incident contact.
Capture the live placement, including account, caption, disclosure, destination, and date. Keep an expiry reminder and takedown owner. Reopen the packet for a new market or channel.
When approval ends, identify who pauses media, removes placements, blocks new exports, and handles stored sources. This prevents an old asset from returning without review.
Summary
A responsible talking photo process separates image authority from subject permission, defines the exact portrayal, gives the client a concrete decision packet, and stops elevated uses with clear hold codes. Public availability is not approval. Release only the version that matches the approved script, voice, context, channel, term, and disclosure, then save proof beside the published file.
Frequently Asked Questions
Does a Client Email Count as Consent for a Talking Photo?
It can be evidence, but it may not answer every required question. Confirm that the sender can authorize the image and the person, then record the approved script, voice, context, channel, language, territory, term, and disclosure.
Can We Animate a Public Figure for Commentary?
Do not assume that commentary removes every risk. Separate editorial purpose from advertising or endorsement, assess whether viewers could be misled, review source rights and platform rules, and escalate the complete concept before production.
Is a Model Release for a Photo Shoot Enough?
Only if its actual scope covers the planned synthetic performance and distribution. Check alteration rights, animated speech, voice, commercial context, channels, term, territory, translations, and reuse instead of relying on the document title.
Should the Subject Approve the Final Export?
When the agreed process requires it, send the actual export candidate with its placement preview. A script approval cannot reveal facial motion, voice delivery, captions, crop, disclosure, or contextual changes that appear in the finished version.
What Changes Require New Approval?
Treat a new script, voice, language, endorsement, emotional tone, crop, campaign, account, channel, territory, term, or disclosure decision as a scope change. Return it to the responsible approver before publishing.
Use the two proof lanes on one real request, then create the approved talking photo in Pippit only after every hold is closed.