Log inStart creating

Seedream 5.0 guide for characters and first frames

How to use Seedream 5.0 Lite, Pro and Flash to design consistent characters and first frames for AI video, with a reference sheet plan and prompts to copy.

, 4 min read

What Seedream 5.0 is for

Seedream 5.0 is an image model, offered in ClickPop in three versions: Lite, Pro and Flash. You use it to draw two things for video: character reference photos and first frames. It does not make video itself; it makes the stills that video models animate.

Those stills do a lot of work. A clear set of character references keeps a face and outfit the same across a whole series. A well-composed first frame sets the light, color and layout of a clip before you spend credits on motion.

Lite, Pro and Flash

Three versions of one model let you match the tier to the task. A practical way to use them is to sketch on one tier and finish on another, keeping the prompt the same so only the level of detail changes.

ClickPop shows the credit price of each before you generate. Run one prompt on all three early in a project, compare what each gives you against its price, and decide which tier you use for exploring and which for final frames. Repeat the test when a version updates.

Once you have chosen a version for a character, stay with it. Switching versions halfway through a series is one of the easiest ways to make a character look slightly different from one episode to the next.

Designing a character in ClickPop

Characters is where you design a fictional adult character once and reuse them everywhere. You choose the image model, Seedream 5.0 or another, and draw 1 to 10 reference photos. ClickPop sends those photos with every video so the face and outfit stay the same.

Here is the order to work in:

  • Write a character description: age range, face, hair, build, outfit and one or two distinctive details.
  • Draw a front view and refine it until you are happy. This is the anchor image.
  • Draw the three-quarter view, then the profile, using the same description word for word.
  • Draw a full-body view so the outfit and proportions are fixed.
  • Add close-ups for expressions you will need, like smiling, serious or surprised.
  • Check all photos side by side and redraw any that drift from the anchor.

Worked example: a character description

Copy this structure and change the details for your own character:

"Theo, a fictional man in his late twenties. Warm brown skin, short tight curls with a faded side part, a neatly trimmed beard, dark brown eyes, slim athletic build. Wears a rust-orange overshirt over a white t-shirt, dark green chinos, white canvas sneakers. A small silver ring on his right index finger. Soft natural daylight, plain light-grey studio background."

Try this prompt in Create Video

Use this exact text for every reference photo and add only the view: "front view, head and shoulders", "three-quarter view facing left", "full body, standing, arms relaxed", "profile facing right", "close-up, smiling".

The plain background matters. Busy backgrounds in reference photos can leak into videos and confuse the model about what belongs to the character.

Keep the description specific but short enough to repeat exactly. Details that matter on screen, like hair shape, outfit colors and one accessory, are worth naming. Details nobody will see in a video, like the brand of a watch, are not, and real brand names should stay out of the prompt anyway.

Drawing first frames

For a first frame, write the prompt as the opening moment of the shot, not as a finished picture. Frame for the motion that will follow. If the camera will push in, start wider. If the character will walk out of frame to the right, place them on the left.

Match the aspect ratio to the final video. Draw vertical frames for shorts and reels, horizontal for YouTube and websites.

For example: "Vertical frame. Theo sits at a small round cafe table by a window, hands around a ceramic mug, looking out at a rainy street. Warm interior light, cool blue light from outside, shallow depth of field. Theo in the lower half of the frame." Then animate it with a motion prompt such as "He lifts the mug, takes a sip and smiles at something outside."

Try this prompt in Create Video

Storyboard mode in Create Video draws stills for each shot first, so you can approve a whole sequence of first frames before animating any of them.

Testing it against other image models

ClickPop also offers Nano Banana Pro, GPT Image 2 and FLUX.2 for characters and first frames. Seedream 5.0 is not automatically the right choice for every project, so run your character description through two or three of them.

Judge by what happens in video, not by the still alone. Animate the front-view anchor from each model with the same motion prompt and see which face holds up through a head turn and a smile. That is the test that matters for a series.

Mistakes to avoid

  • Changing the description between reference photos. Paste the same text every time.
  • Busy backgrounds in reference photos. Use a plain studio backdrop.
  • Too few angles. A front view alone leaves the profile to chance.
  • Switching Seedream versions or image models mid-series.
  • Designing a character that resembles a real person or using real logos on clothing. Keep characters fictional, adult and unbranded.

Questions

Is Seedream 5.0 a video model?

No. It is an image model you use for character reference photos and first frames, which video models then animate.

How many reference photos should a character have?

ClickPop's Characters studio takes 1 to 10. Front, three-quarter, profile and full body cover most needs, plus close-ups for key expressions.

Which Seedream 5.0 version should I use?

Run one prompt on Lite, Pro and Flash, compare results with the price shown before each render, then pick one tier for exploring and one for finals.

Seedream 5.0 guide for characters and first frames | Click Pop