Website URL or Product Photos? Choosing the Right Brand Reference for AI Marketing Videos

TL;DR

Use a website URL only when the video tool can actually read it and you need brand or product context. Use product photos when the shot needs to show a particular product. A URL is not a visual reference, and a photo is not a verified source for product claims. If both the message and the product’s appearance matter, provide both where supported, then test one simple shot and inspect the result before making a full sequence.

Key takeaways

• Choose a reference based on the main risk: an incorrect claim calls for verified product information; an unrecognizable product calls for a clear product image.
• Check whether the specific tool can read URLs or accept image references. Don’t assume that including a link means the tool has read the page.
• Give references distinct jobs: approved notes or readable website content for context, and photos for visual direction.
• Start with one subject and one clear action. Change one input at a time when comparing results.
• Review generated claims, logos, packaging, and text before publishing.

What a brand reference needs to do

A brand reference is useful when it answers a specific creative question. What does the product look like? Which approved benefit should the video communicate? What tone should it use? Those are different jobs, and one reference may not cover them all.

For a marketing video, the brief should make the video type, duration, brand name or website, features or topic, and call to action clear. InVideo’s guidance also warns that vague prompts can lead to generic visuals, while more specific direction gives the generator a clearer task. (https://invideo.io/blog/ultimate-ai-prompting-guide-for-marketing-videos)

The practical distinction is context versus appearance. Product-page copy or approved notes can tell you what the product does and how the brand describes it. A product photo can show its shape, color, label placement, or finish. Treat these as inputs to guide a generation, not as guarantees that the final video will reproduce every detail correctly.

Website URLs vs. product photos

Use a URL for context, if the tool can read it

A website URL can be a useful starting point when you need to draw on product information, positioning, or brand language. But URL handling depends on the particular tool and interface. Check whether it offers website reading or importing. If it doesn’t, pasting a link may not give the generator the page’s contents.

When the tool cannot read the page, copy the relevant approved information into your prompt instead. Keep it focused: one benefit, the intended audience, and any wording or claims that must not change. Check generated claims against approved materials before using the video. A link—or a generated sentence that sounds plausible—is not a substitute for that review.

Use a product photo for visual direction, if references are supported

A product photo is the more direct reference when the viewer needs to recognize a specific item. It can guide a shot of a package on a counter, a product in someone’s hand, or an item turning in place. For image-to-video prompts, FlexClip recommends specifying the subject, action, background, background movement, and camera movement. (https://help.flexclip.com/en/articles/10326783-how-to-write-effective-text-prompts-to-generate-ai-videos)

A photo is still guidance, not a guarantee of exact reproduction. Inspect the result for changed colors, warped logos, altered labels, or invented text. If exact legal copy or packaging text matters, plan to verify it and make any necessary corrections after generation rather than assuming the model will preserve it.

Use both when both jobs matter

For a product demo, you may need the generator to understand what the product is for and what it looks like. If the tool supports both kinds of input, use the website or verified notes for context and product photos for appearance. State those roles clearly in the prompt. Neither reference should be expected to correct an error in the other.

Video task	Best starting reference	Prompt focus	Check the result for
Product demo	Clear product photo	One product action and a simple setting	Shape, label, logo, and whether the action makes sense
Lifestyle clip	Product photo plus approved brand notes	Product, audience, setting, and mood	Product consistency and whether the scene fits the message
Brand-awareness video	Readable website content or a written brand brief	One approved takeaway and a clear call to action	Claims, tone, campaign message, and generated text

A useful decision rule: start with the input that addresses the most costly mistake. If the video could misstate a product benefit, begin with approved facts. If viewers might not recognize the product, begin with a clear image. If both outcomes matter, test both inputs together after confirming the tool supports them.

A worked example: the same bottle, two tests

Suppose you’re creating a short video for a reusable water bottle. Its approved product information says it is insulated, and a product photo shows its shape and label. You want a simple morning-routine shot, not a sequence of scenes.

Weak prompt: “Make a cool video for our bottle.”

Context-led version: “Create a concise brand-awareness shot for a reusable water bottle. Use this approved benefit: insulation. Show a calm morning routine with a practical, confident tone. Do not add performance claims or on-screen text.” If the interface can read the product page, provide it as context. If not, paste the verified benefit and approved voice notes directly into the brief.

Image-led version: Attach the product photo if image references are supported, then write: “The same bottle stands on a kitchen counter as a hand picks it up. Soft morning window light; gentle camera push-in; uncluttered scene.” The prompt gives the product one clear action and specifies the setting and camera movement, following the kind of image-to-video structure described by FlexClip. (https://help.flexclip.com/en/articles/10326783-how-to-write-effective-text-prompts-to-generate-ai-videos)

Compare the outputs against separate questions: Does the context-led version communicate the approved benefit without adding a new claim? Does the image-led version retain the bottle’s recognizable appearance? If the label is wrong, simplify the shot or adjust one prompt element; don’t simultaneously change the image, setting, action, and wording. A controlled comparison helps you decide what to change next.

For richer scenes, make the composition explicit rather than piling on adjectives. Kling’s prompting guide recommends describing background, middle-ground, and foreground layers, and including environmental forces such as wind or rain when they affect the scene. (https://kling.ai/blog/ai-video-prompts-advanced-techniques-realistic)

If you want to run that image-led test without building a separate audio step, Imaginary offers image-to-video generation with Google's Veo 3.1, whose generations include synchronized audio in a single pass. Imaginary is an independent platform and is not affiliated with or endorsed by Google; the model name describes which model generates the video.

A reusable brand-reference brief

Copy this template and fill it in before generating. Include a URL only if the interface can read it. Otherwise, replace it with verified notes. Attach an image only if image references are supported.

Goal: [product demo / lifestyle / brand awareness]

Audience and takeaway: [who it is for; one approved message]

Verified product facts: [claims approved for this video]

Brand voice: [three concise traits or approved wording]

Visual reference: [attached image; details it should guide]

Shot: [one subject, one action, setting, camera movement]

Do not invent: [claims, features, prices, logos, or text]

Review before use: [appearance, claims, logo, packaging, generated text]

Keep the opening of the prompt focused on the most important information. DeepReel recommends putting key information first and gives a subject, style, camera, mood, and duration formula for video prompts. (https://deepreel.com/blog/ai-video-prompts)

Checklist

• Confirm whether the tool can read website URLs and accept image references.
• Choose the input that addresses the biggest risk: wrong context or wrong appearance.
• Copy verified product facts and approved brand wording into the brief.
• Give the image and the website or notes distinct jobs if you use both.
• Describe one clear action, a specific setting, and any necessary camera movement.
• Generate one short test before planning a full sequence.
• Change one input at a time when diagnosing a weak result.
• Check product details, claims, logos, packaging, and generated text before publishing.

FAQ

Should I paste my website URL into an AI video prompt?

Only if the tool or interface can read website content. Otherwise, paste the relevant approved product details and brand language into the prompt; a URL alone may not provide context.

Are product photos better than a website URL for an AI product video?

They are a more direct choice when the shot depends on how the product looks and the tool supports image references. A readable website or written brief is more useful for context such as approved benefits and brand voice. Neither guarantees an accurate result.

How do I keep packaging accurate in an AI marketing video?

Provide a clear product image if references are supported, keep the shot simple, and inspect the result closely. Don’t assume that labels, logos, or text will reproduce exactly. If those details must be exact, verify and correct them before publishing.

Can I use a product page and product photos together?

Yes, if the interface supports both. Use the page or verified notes for product context and the photos for visual direction. Then check the generated video against both references.

What should I test first?

Start with one short shot and one clear action. Compare a version with and without the image reference while keeping the prompt otherwise unchanged. That makes it easier to see whether the reference improved product recognition.

Turn the test into a repeatable workflow

A small comparison can help you choose the right reference before it shapes an entire campaign. Imaginary is an independent, multi-model AI video generator with one account and one credit balance for several video models. It offers Google's Veo 3.1 in Fast, Quality, and Lite modes, as well as text-to-video and image-to-video generation. New users get free credits to start, so you can try a reference-led shot and review the result. Imaginary is not affiliated with or endorsed by Google.

When your brief is ready, use Imaginary to test a product image against your approved context and find a generation approach that fits the shot. For repeatable workflows, Imaginary also offers a developer API for generating videos programmatically.

Start creating with Imaginary

Imaginary lets you generate AI video with native audio, up to 4K, from one account with credits that never expire — and free credits to start. Veo 3.1 is live now, with Seedance 2.5 and Wan 3.0 coming soon.

• Generate your first video now: Get started
• Compare plans and pay-as-you-go pricing: See pricing