How to Create Sanjay Dutt AI Videos in 2026 (Step-by-Step Guide)

If you have spent more than five minutes on Instagram Reels or TikTok lately, you have probably seen the trend. Classic 90s Bollywood nostalgia is back with a vengeance, and “Baba” himself—Sanjay Dutt—is front and center. I kept seeing these insanely high-quality, cinematic AI videos of him walking in slow motion, wearing his iconic tinted sunglasses, looking like absolute royalty.

At first, I thought people were using some massive, expensive workflow. A few months ago, making these meant generating an image in one app, porting it over to another complicated platform to animate it, and hoping the face didn’t warp into a melted candle. It was exhausting.

But I’ve spent the last few weeks testing out the new tools we have access to right now in 2026, and I realized something: you can do the entire process, from scratch, straight inside Gemini.

Google completely overhauled its generation models this year. With the integration of their new image model (Nano Banana 2) and their heavy-duty video model (Veo), everything is baked right into the chat interface. You don’t need to be a coder or a VFX artist to make this work. You just need to know the right buttons to press.

Since you already have your own secret sauce—your specific image and video prompts—I am going to skip the prompt-writing theory. Instead, I am going to walk you through my exact personal workflow for taking your prompts and turning them into incredibly realistic, trend-ready Sanjay Dutt videos using Gemini.

Grab a coffee, open up your browser, and let’s get into it.

What You Need Before We Start

Before we dive into the steps, let’s get our workspace sorted. I highly recommend doing this on a desktop or a laptop rather than your phone. While the Gemini app is great, having a bigger screen makes it much easier to review the fine details of the images and videos you generate.

Here is your checklist:

• A Gemini Paid Subscription: This is the one non-negotiable. To access the Veo video generation model and the high-fidelity Nano Banana 2 image tools, you need to be on the Gemini Pro or Ultra tier. The basic free tier won’t give you the video capabilities you need for this.

• Your Custom Prompts: Have your text file open with the image prompt and video prompt you already created.

• Patience: AI video generation takes a minute. Even with state-of-the-art servers, rendering high-quality video frames requires a little waiting time.

Phase 1: Generating the Perfect Base Image

The golden rule of AI video is that your video is only ever as good as your starting image. If the base image looks slightly off, the video will amplify those mistakes tenfold. We are going to use Gemini’s image generation tools to lock in that classic Sanjay Dutt look first.

Step 1: Fire up the Chat

Open up Gemini on your browser. Make sure you are logged into your paid account. Start a completely fresh chat. I always recommend starting a new chat for a new project so the AI doesn’t get confused by previous conversations.

Step 2: Input Your Image Prompt

Go to the chat box. You don’t need to overcomplicate the instructions to the AI. Simply tell it that you want to generate an image and then paste your specific prompt.

It should look exactly like this:

Step 3: Review and Select

Hit enter. Gemini is currently using the Nano Banana 2 model under the hood, which is incredibly fast. Within a few seconds, you will get a batch of images to look at.

• Pro Tip: Look specifically at the eyes and the jawline. Sanjay Dutt has very distinct facial features. If the AI messes up the eyes, don’t use that image.

• If you are on an AI Plus, Pro, or Ultra plan and you want even more cinematic detail, you can click the three-dot menu on your favorite image and select “Redo with Pro.” This runs it through an even heavier refinement process. I usually do this for my final selections just to get that extra crispness in the skin texture and lighting.

Step 4: Lock It In

Once you have the perfect image of Baba, right-click and download it to your computer, or just keep it active in the chat. We are going to use this exact image to drive our video.

Phase 2: Bringing Baba to Life (The Video Process)

This is where the magic happens. We are going to transition from the image tools over to Gemini’s Veo video model. Veo has a feature that allows you to use a reference image to guide the video content. This means we are forcing the AI to keep our perfect Sanjay Dutt face and style consistent while adding motion.

A quick warning about quotas: If you are on the Pro tier, you have a strict limit of 3 video generations per day (5 if you are on Ultra). Do not waste these. Make sure you are completely happy with your base image before moving to this step.

Step 1: Upload the Reference Image

In the same Gemini chat, click the plus (+) icon next to the text box to upload a file. Select the perfect image you just generated (or downloaded) in Phase 1.

Step 2: Apply Your Video Prompt

Now, we are going to tell Gemini’s Veo model to animate it based on your instructions. You need to explicitly tell Gemini to use the attached image as a reference for the video.

Type this into the chat box along with your uploaded image:

Step 3: The Waiting Game

Hit generate. Because Veo is building high-fidelity video frames, this will take longer than the image generation. Grab some water, stretch your legs. It usually takes a few minutes.

Step 4: Download and Inspect

Once the video is done, hit play. Watch the motion. You are looking for a smooth transition. Veo is fantastic at understanding lighting changes and camera pans. If the movement matches your prompt and the face stays consistent, you have struck gold. Download the video file to your hard drive.

Common Mistakes I Made (So You Can Avoid Them)

When I first started trying to animate character-specific videos like this, I burned through my daily quotas by making silly mistakes. Here is what you need to watch out for:

1. Too Much Motion in the Prompt

Even in 2026, pushing an AI video model to do too much at once will break it. If your custom video prompt asks for the character to run, jump over a car, and change clothes, the face will warp, and it won’t look like Sanjay Dutt anymore. The best AI videos right now rely on subtle, cinematic movements. Think slow-motion walks, smoke exhales, or slow camera zoom-ins.

2. Forgetting the Aspect Ratio

If you are planning to post this on TikTok or Instagram Reels, you need a vertical video (9:16). If your initial image prompt generated a widescreen (16:9) image, animating it and then cropping it later on your phone is going to ruin the resolution. Make sure your base image is already in a vertical format before you send it to the video model.

3. Ignoring the Lighting Consistency

Sometimes, the video model tries to add its own dramatic lighting that clashes with the base image. If your base image is set in a dark, moody alleyway, make sure your video prompt doesn’t ask for “bright, sunny daylight motion.” Keep the environment consistent between both prompts.

Real-World Use Cases: What is Working Right Now

You might be wondering how people are actually packaging these videos once they download them from Gemini. From what I am seeing across social media, there are a few formats that are practically guaranteed to get traction:

• The “Khalnayak” Retro Edits: People are taking these high-def AI clips of Baba, bringing them into editing apps like CapCut, dropping a heavy bass boosted 90s Bollywood track over them, and adding a slight VHS grain filter. It bridges the gap between modern AI tech and raw nostalgia.

• Alternate Universe Concepts: Fan pages are huge right now. Using this Gemini workflow, I have seen creators make videos of Sanjay Dutt cast in movies he was never in—like placing him in a cyberpunk setting or a Peaky Blinders-style mafia world.

• Motivational Shorts: It sounds crazy, but a slow-motion, intense video of Sanjay Dutt smoking a cigar, overlaid with a deep, AI-generated voice reading motivational quotes, is pulling millions of views on YouTube Shorts.

Wrapping Up the Workflow

Making high-quality AI videos used to feel like learning a programming language. Now, with Gemini integrating both the Nano Banana 2 image tools and the Veo video model into one seamless interface, the barrier to entry is basically gone.

The process is straightforward: respect your daily quotas, make absolutely sure your base image is flawless before you try to animate it, and keep your motion prompts focused and cinematic. Since you already have your custom prompts dialed in, you are already past the hardest part. Just drop them into the workflow above, and you will have a viral-ready clip in under ten minutes.

Drop a comment below if you run into any weird rendering glitches, and happy editing!