APOB coffee-host workflow: a screenshot user guide

Publisher: APOB AI. Observed September 15, 2026. This guide documents an APOB-only walkthrough and account-specific limits.

How to try APOB before comparing it with Pollo

This walkthrough follows one fictional coffee host from reference image to a scene and a short clip. Screenshots show the actual desktop app on September 15, 2026. The account displayed Nano but had extra model allowances and no corresponding generation debits in the inspected history; use the quoted costs as examples, not a promise of ordinary free-account entitlements.

1. Open the official app and check the trial conditions

Open app.apob.ai and sign in if prompted. This session was already signed in. Open Upgrade for the current offer; Profile → Usage details shows your balance and refill, while Credit usage table records spending and returned credits.

For a coffee-host concept, use a fictional person and nonconfidential assets. The public Nano offer showed 80 credits/day, watermarked output, no private models/content and one retained portrait model. Check eligibility before creating another model; an account-specific allowance is not the public plan. Billing help

Step 1: inspect the current offer before treating the free allowance as a production budget.

2. Create and save a fictional host reference

Choose Create Portrait Model. Upload a rights-cleared face, or open AI influencer generator. We selected Man, left the age at 25, entered a fictional coffee-educator description in Bespoke Details, and generated one reference at a displayed quote of 10 credits. Select the completed result with Select this image as reference image.

Name the model—ours is Coffee Host Guide Sep15—and check Public before clicking Create your image model now. Wait until training finishes. The reference unexpectedly used a car interior rather than our requested beige background; its face was usable for this workflow demonstration, but the scene instruction was not fully followed. Choose a clearer reference if your task needs a clean backdrop. Portrait-model help

Step 2: reference selection and model creation are separate actions. This form was revisited after creation for a clear capture; it was not submitted twice.

3. Insert the model and set the image batch

In the model's workstation, open Generate image → Chat to generate. Write the kitchen scene, then click the model control above Description to insert the real named model chip into the text. The example chip in an empty editor is only a placeholder. In our submitted prompt, the actual chip appears after the scene text.

Select 9:16 and Fast. The interface quoted 20 credits for four images and showed one Element slot. Fast had no Quantity selector; higher image tiers expose different quantity and reference options. Our grinder was described only in words—no product Element was attached. For a real SKU, attach its approved reference and inspect the result; a text description cannot establish product fidelity. Quantity rules · Model and Element references

Step 3: check the chip inside the submitted prompt and the total beside Generate.

4. Inspect every result before choosing one

Click Generate, wait for the Generation feed to finish, and inspect the entire batch. We received four 576 × 1024 images. All four depicted a coffee host, but the requested small manual grinder became larger countertop equipment. None met that product-detail requirement. Keeping this failure visible is more useful than showing only an attractive face.

Compare the face, hands, outfit and product separately. For this walkthrough, image 1 was retained to demonstrate the next control—not approved as an accurate grinder advertisement. For production, correct the still or supply the correct product reference before animating it.

Step 4: a completed generation is not automatically an accepted deliverable. Inspect all four originals.

5. Open the result and verify the exported image

Open a completed image. The full-screen preview has a Download arrow; the standalone result page puts Download under More actions. We exported all four JPEGs through Chrome after the in-app browser's download action did not produce a retained file. The files opened successfully and measured 576 × 1024.

The inspected JPEGs had no visible watermark, despite the Nano label. Because this account also showed nonstandard allowances, these files do not establish normal Nano watermark behavior. Follow the published free-plan rules and test your own export before promising a clean client deliverable. Download and watermark rules

Step 5: retain the downloaded file and result link, not just a preview screenshot. Image 1 result.

6. Send the still to Image to video and check the actual file

From the image preview, choose Image to video; the selected still becomes the first frame. Describe one modest movement, then recheck Quality, Resolution, Duration and the total quote. Our settings were Fast · 480P · 4s, with Native audio visibly enabled. Add last frame and Add shot were inactive in this configuration. The quote was 8 credits/s, 32 credits total.

We generated one clip using a gentle camera push-in. We saved the MP4 from the generated media resource; the app’s video-download path and its export watermark behavior were not tested. The retained video resource measured 480 × 864, with a 4.458-second video track and an audio track present. We inspected frames at 0, 2 and 4 seconds; this was not a lip-sync or supplied-audio test. Check playback and the actual dimensions, duration and audio before preparing a platform upload. Native-audio labels alone do not prove the sound you need is present. Video controls

Step 6: the source image still contains the wrong product. Inspect the original clip and frames; use it as a workflow example, not product-fidelity proof.

To compare Pollo fairly, retain the same rights-cleared inputs, deliverable requirements and attempt budget. The APOB walkthrough above does not replace a matched run on both services.

FAQs about APOB AI vs Pollo AI

Is APOB AI free to try, and do the credits expire?

The observed public Nano offer is $0 with 80 credits/day and one-day credit validity; the homepage says no card is required. Check the expiry of your actual grant in Profile. Do not assume you can save several days of free credits for one larger task. APOB · Credit grants and expiry

Does Pollo offer free credits or recurring rewards?

Its observed pricing table showed 20 free credits. The official FAQ lists signup, daily check-ins and referrals among bonus sources. Account-specific amounts and claim schedules were not verified; this page does not describe Pollo as a one-time-only trial. Pollo credits

Where should I sign in and compare the real prices?

Use the APOB app and APOB AI pricing, or Pollo sign-in and Pollo AI pricing. Separate month-to-month prices, first-month promotions and annual monthly equivalents. Confirm the full commitment, taxes and renewal total before payment; a competitor page about APOB is not APOB's login service.

Which service costs less for an accepted result?

That remains unmeasured. Our APOB walkthrough displayed quotes of 10 credits for the reference, 20 for four scene images and 32 for the clip—62 quoted generation credits in total. Model creation showed no separate quote, and corresponding debits were not visible in the inspected history. This is neither measured spend nor a complete ordinary-account cost. Compare all preparation, rejected versions, editing and export costs per accepted deliverable.

Can I get watermark-free downloads by upgrading later?

APOB's official rule says earlier free generations retain their watermark after an upgrade; watermark-free production requires the relevant paid entitlement when generating. Pollo documents an active-subscription condition for watermark-free downloads. Our nonstandard account's unwatermarked JPEGs do not override those rules. APOB downloads · Pollo downloads

Can I use generated content in ads or client work?

Check plan rights and current terms, and obtain permission for all reference images, voices and likenesses. Provider terms about generated-content ownership do not guarantee copyright protection, exclusivity or a person's endorsement. A product ad must also accurately depict the real product. APOB terms · Pollo terms

Which platform keeps a recurring character more consistent?

No cross-platform winner has been established. APOB's saved portrait model can be reused through an actual model chip in the composer. Our one kitchen batch is not a three-location identity benchmark. Judge complete scene sets on both products, separating face, clothing, body proportions and product details.

Why is Create your image model now unavailable?

Check that you are signed in, a reference image has finished loading, and the required Model name is filled in. The help center specifies a 32-character name limit. Model allowances also depend on the plan; do not delete an existing character simply to follow a tutorial. Wait for model training to finish before generating with it. Portrait setup

Why does Fast create four images when there is no Quantity control?

Fast returns a four-image batch. Higher image tiers expose Quantity choices, usually one or two; more outputs change the quote. The live Generate total is the useful number to record, rather than assuming every tier charges for a single image. Aspect ratio and quantity

Why do model references or Elements disappear when I change quality?

Reference allowances vary by mode and quality. The tested Fast image composer showed one Element slot. If a lower tier supports fewer references, APOB can ask to remove the excess Elements and their mentions. Read that confirmation and inspect the actual chips before generating; typing a reference name as ordinary text does not attach it. Composer reference rules

Can I combine a last frame with multiple video shots?

APOB documents these as incompatible options: adding a last frame disables multi-shot, and multi-shot excludes a last frame. Last-frame availability also depends on quality. In our Fast 480P setup, both Add last frame and Add shot were inactive. Recheck the controls after every quality change. Image to Video

Is Native audio the same as a talking-avatar or lip-sync workflow?

No. Native audio generates sound with the clip. A talking-avatar workflow needs an image and an audio track; lip sync needs a video and audio, subject to input limits. For the spoken coffee tutorial, preserve the same authorized recording on both services instead of substituting unrelated generated speech. Our clip had an audio track, but no uploaded recording was tested. Input requirements · Pollo avatar

Can I keep an unreleased product or private portrait confidential?

The public APOB Nano offer excludes private models/content. Pollo documents an active-subscription requirement for community-sharing opt-out. Private display is separate from provider retention and contractual confidentiality; inspect both before uploading client materials. APOB visibility rules · Pollo sharing rules

What should I do if generation fails or the product looks wrong?

A failed processing task and an incorrect completed picture are different cases. APOB's help center says credits from failed generations are returned automatically; verify the returned entry before counting a refund. For an incorrect product, inspect the source, attach the correct reference or revise the still before making a video. A more expensive tier is not a guaranteed fix. Failure help

Why does a video preview appear silent?

Browser autoplay often starts muted. Use the speaker control when present, then inspect the downloaded file's audio track and test another player. An audio track's presence does not establish correct speech or lip sync. If playback remains broken, retain the result link, settings and credit record for support. Preview and sound controls

Will the files work for TikTok, Reels and YouTube Shorts?

Check actual dimensions, aspect ratio, duration, format, audio, captions and watermark against the destination's current requirements. Our 4s setting produced a video track of about 4.46 seconds, so inspect the export instead of relying on the label alone. Apply relevant AI/ad disclosures; no direct publishing integration or automatic platform approval is established here.

Exact inputs used in this walkthrough

These preserve what was actually submitted, including instructions that the outputs did not fully follow. They are examples to inspect, not validated product-ad recipes.

Reference portrait

Enter in AI influencer generator → Bespoke Details; Man, age 25. The generated background differed from the request.

An original fictional adult home-coffee educator, age 25, short wavy dark-brown hair, brown eyes, clean-shaven, plain olive-green cotton shirt. Friendly neutral expression. Head-and-shoulders portrait facing the camera, soft natural window light, simple warm beige background. No celebrity resemblance, no logos, no text.

Kitchen scene

Enter in Chat to generate and separately insert the real Coffee Host Guide Sep15 model chip. Fast, 9:16, four images. No grinder Element was attached; product details failed.

A vertical editorial photograph of an adult coffee educator in a home kitchen. He wears a plain olive-green cotton shirt and stands behind a light oak counter. A small matte-black manual coffee grinder with a silver crank rests upright on the counter beside a white ceramic cup. Both hands rest naturally on the counter, away from the grinder. Soft morning window light, eye-level medium shot, realistic proportions. No logos, captions, invented labels, extra fingers or duplicate objects.

Motion description

Enter in Image to video → Description with image 1 attached. Fast, 480P, 4s; Native audio remained enabled.

A slow, gentle camera push-in toward the adult coffee host in this kitchen. Keep his face, olive-green shirt, hands, cup, countertop and coffee equipment in their original positions and shapes. The host maintains a calm expression with only a natural blink. No hand gestures, object movement, scene cuts, subtitles, logos or added objects.