The AI tools we use
No single model wins every job. Each one has its strengths, and they change fast: several of the models below released a major new version during 2026. So we choose per shot, not per project.
Tool | Made by | What we use it for |
|---|---|---|
Higgsfield | Higgsfield AI | Our main workspace. It brings many image and video models into one place, with tools for character consistency, camera control and upscaling |
GPT Image | OpenAI | Product and editorial stills that follow detailed prompts and reference photos closely |
Nano Banana Pro | Fast image creation and editing, often to build the first frame of a video | |
Seedance 2.5 | ByteDance | Brand films built from reference images, with longer takes and edits to a single moment |
Kling | Kuaishou | Turning a still image into smooth video with controlled camera moves |
Veo 3.1 | Landscapes, drone-style shots and natural movement such as leaves and water |
We wrote a closer look at one of these models in Finally, an AI video you can edit instead of regenerate. AI also reaches beyond visuals: the booking app for Pattaya Prestige Vans, also on our Experiments page, was built entirely with AI on Replit.
How an AI production actually runs
The button is the easy part. This is the process behind a result worth publishing.
Find the visual gap. Before generating anything, we study the brand’s current images and the brands it admires. For Gango, a kitchen knife brand, the gap was clear: the banners showed knives as catalogue objects. The new direction showed moments in a real kitchen, where the knife is one element of the frame.
Build a reference pack. We use two kinds of reference. An anchor image of the real product keeps its exact shape and details. Style references set the light, mood and composition. For Gango, the anchor kept the knife’s silhouette and brass rivets accurate.
Write the prompt like a shot list. Subject, surface, props, light direction, colour temperature, camera angle and palette, plus a clear list of what must not appear.
Approve stills first. Most videos start as still frames. Fixing a frame is far quicker than fixing a moving clip.
Animate and test in drafts. Approved frames become short clips. We review low-resolution drafts and compare versions side by side, for example a text-only prompt against one driven by real reference images.
Select, fix and finish. We keep the best takes, correct small details, add brand typography in editing rather than inside the AI, then cut the clips together with sound.
What drives the cost of AI content
Many people assume AI content is almost free because the machine does the work. In practice, the AI credits are the smaller part of the cost. Most of it is human time: preparing references, prompting, choosing results and fixing what is almost right. Five things decide how much time a project needs:
Complexity. A product on a plain surface is quick. A scene with hands, food and moving light takes more attempts.
Accuracy. The more exact the product must look, the more generations it takes to get it right.
Revisions. One clear, consolidated round of feedback keeps the number of rounds down.
Volume. Visuals produced together in one batch cost less each, because references and style are set once. A monthly content plan from our social media marketing team is the easiest way to work in batches.
Video length. A video is several short AI clips edited together, so a longer film means more clips to generate, match and cut.
When you compare quotes, ask how much is AI credits and how much is people’s time, and what happens if the first results miss.
What brands use AI images and video for
AI content creation works best where a brand needs many visuals, fast, in many formats. These are the uses we see most:
Social media content. Regular posts, Stories and Reels in every aspect ratio, built from the same reference pack so the feed stays consistent. This is where AI saves the most time for our social media marketing clients.
Ad variations. Paid campaigns need several versions of each visual to test what works. AI makes it practical to produce those variations for Google Ads display and video campaigns, as well as Meta and TikTok, alongside our advertising design work.
Website hero videos and imagery. A looping hero video or a set of lifestyle images can lift a homepage or a landing page without a full shoot.
Packaging and print previews. AI mockups show new packaging design or a brochure in real settings before anything is printed, which helps decisions and pre-launch marketing.
Launches before the product exists. When a product is still in production, AI visuals built from renders or samples let the marketing start on time.
Press and PR visuals. Campaign images and short clips for announcements support digital PR when there is no budget or time for a dedicated shoot.
What AI does well: examples from our work
Our Experiments page shows client projects and internal tests side by side. A few patterns stand out.
Hero videos for websites. For Mekong Tourism, we generated the introduction video that runs as the hero banner on their new website. A travel film that would normally need a crew and a location shoot became a production task at the desk. It pairs naturally with our web design and development work.
Social video for B2B brands. We produced AI video for the social channels of Hitachi Construction Machinery Thailand. Heavy machinery is hard to film on demand, and AI lets the brand post regularly without booking a site visit for every clip.
Product and lifestyle imagery. For Gango, we created editorial hero images for the website. For a jewellery brand, we generated an image series showing gemstone polishing up close. For Lia Coffee, we produced mockups that present the branding we designed in real settings, which is useful when packaging and signage are not printed yet.
UGC-style content. Our Bangkok Tea test shows a customer enjoying the product in an everyday setting, the casual style that performs well on TikTok and Instagram.
Mixing AI with real material. For Invidi, we combined AI-generated footage with existing material to introduce a new service. This hybrid approach often gives the best of both: real brand assets where accuracy matters, AI where imagination helps.
Keeping AI content on brand
The biggest risk with AI content is not poor quality. It is visuals that look good but do not look like your brand. A feed full of beautiful but unrelated images weakens recognition instead of building it.
Consistency comes from decisions made before the first prompt: a defined colour palette, a photographic style, rules for how the product is shown, and a clear tone. This is why AI production works best on top of a clear brand strategy and identity system. If those rules are written down, as we describe in our overview of corporate identity and its 18 key elements, they can be turned directly into prompts and reference packs.
Three practical rules we follow:
Logos and brand type go in during editing. AI models still misspell words and distort logos. Brand typography is added afterwards, so it is always correct.
One reference pack per brand. The same anchors and style references are reused across every project, so images made months apart still belong together.
A human checks every frame. Hands, product details, reflections and text in the background are where AI errors hide.
Rights, transparency and responsible use
AI content raises fair questions about ownership and trust. Our approach is simple:
Use your own references. We build from the client’s own product photos and assets, and use other brands’ visuals only as mood references, never as material to copy.
No real people without consent. We do not generate recognisable real people. Founders, staff and ambassadors are photographed or filmed for real.
Respect platform rules. Major AI platforms add content credentials and limit what can be generated from real faces. Platform permission still does not replace proper rights clearance.
Protect what you create. Names, logos and key visuals should be protected like any other brand asset. Our guide to protecting your brand in Thailand covers the legal side.
Where we still recommend a real shoot
AI is a strong tool, not a replacement for every camera. We recommend real photography or video when:
Exact product detail matters. Packaging text, typography, watch faces and jewellery geometry must be perfect. AI can come close, but close is not good enough for an e-commerce product page.
Real people are involved. Founders, staff and brand ambassadors should be filmed or photographed for real. Generating a known person raises consent and rights issues.
The deadline has no room for retries. AI results vary between attempts. For a launch date that cannot move, a planned shoot with a fallback is safer.
Often the best answer is both: a short real shoot for hero product shots and people, then AI to extend those assets into dozens of formats and scenes.
How to brief an AI content project
A good brief cuts production time, and production time is most of the cost. Bring us:
Reference images. Product photos from several angles, packaging, logo files and any past campaign visuals.
Brand rules. Colours, fonts, tone and anything the brand never does.
Examples you like. Two or three visuals or videos that capture the mood you want, from any brand.
Formats and quantities. How many images or videos, in which aspect ratios, for which channels.
Usage. Organic social, paid ads, website or print, so we deliver the right specs.
Our guide on how to brief a creative agency covers the wider briefing process. For AI work, the reference images matter most: the more the model sees of your real product, the less time we spend correcting it.
AI images, video and SEO
Original visuals also help search. Stock photos appear on thousands of websites, while AI images made for your brand are unique to your pages. To get the SEO benefit:
Write real alt text. Describe what the image shows and include the topic of the page, not a list of keywords.
Compress before upload. AI images and videos are often large files. Heavy pages load slowly and lose rankings, which is why web hosting and maintenance includes image and video optimisation.
Use video where it adds value. A short hero video can keep visitors on a page longer, but it should never block the content from loading.
Name files clearly. A file called “bangkok-coffee-shop-latte.jpg” tells search engines more than “IMG_2481.jpg”.
Our SEO team handles these details on every site we manage, and our SEO guide for Thailand explains the wider picture.
Is AI content right for your brand?
AI content works best for brands that need more visuals than a traditional shoot can supply: regular social posts, ad variations, seasonal campaigns and website imagery. It works best on top of a clear brand, because a model can only stay consistent with a style that is already defined. If your visual identity is not settled yet, start with logo and brand identity work first.
We use AI the same way across the studio: to remove slow, repetitive work so the team can spend more time on ideas and quality. Our web team described the same shift in AI changed how we work, it didn’t replace us.
Browse the AI Experiments page to see the results, then send us your brief for a quote based on your exact project.
FAQ
Which AI tools are best for brand images?
It depends on the shot. We use GPT Image for product and editorial stills that must follow reference photos closely, and Nano Banana Pro for fast creation and editing. Most of our work runs inside Higgsfield, which gives access to many models in one place.
Which AI tools are best for brand video?
Seedance 2.5 for reference-led brand films, Kling for turning still frames into smooth camera moves, and Veo 3.1 for landscapes and natural movement. We often combine several in one project.
Is AI content cheaper than a photo shoot?
Often, especially for large volumes of social or web visuals. But AI credits are only part of the cost. Most of it is the human time to prepare references, prompt, select and fix results.
Can AI replace a product photo shoot?
Not always. When packaging text, typography or exact product details must be perfect, or when real people appear, we recommend a real shoot. Many projects combine a short shoot with AI to extend the assets.
How do you keep AI images consistent with a brand?
With a reference pack: real product photos as anchors, style references for light and mood, and clear brand rules. Brand typography is added in editing rather than generated.
Let’s keep in touch.
Follow Asia Media Studio for design insights and updates on Facebook, Instagram, and LinkedIn.
