Skip to main content
varadatech
varadatech
Participant
April 21, 2026

Video Generation Fails with “Something Went Wrong”

  • April 21, 2026
  • 2 replies
  • 37 views

I am currently using Adobe Firefly under a free trial. Whenever I try to generate a video, the process fails and shows the error message “Something went wrong.” This issue occurs every time, regardless of the prompt used.

Steps to reproduce:

  1. Open Adobe Firefly

  2. Go to the video generation feature

  3. Enter any prompt and start generation

  4. Wait for processing

Expected result:
The video should be generated successfully.

Actual result:
The generation fails with a “Something went wrong” error, and no output is produced.

This issue persists across multiple attempts and prevents me from using the feature during my trial period. Please clarify whether this is a bug, server limitation, or a restriction of the free trial, and provide a fix or guidance.

      2 replies

      O
      Community Expert
      Oh.N8Community Expert
      April 24, 2026

      Hey ​@varadatech 

      1) Rohan" as a character, kingdom, or term from The Lord of the Rings is protected by copyright as part of J.R.R. Tolkien’s literary works, and the name "ROHAN" is registered as an active trademark for various goods by Middle-earth Enterprises (Saul Zaentz Company)

      In cases like this describe the character. What does the character look like, essentially?

      Besides that your prompt might be too detailed or specific. Remember Firefly only makes a 5 second video not a movie.

      You will need to break the prompts into a series of shots/ takes.


      2 use active voice. Ask specifically for what you want. 
      Example: No splash could give you a splash.

      Voice over…you may need to use a combination of apps for a proper voice over.

      Cheers
      Nate
       

      O
      Community Expert
      Oh.N8Community Expert
      April 22, 2026

      Hey ​@varadatech 

      Help us. Help you. 
      Could you provide an example of your prompt? The model being used?

      Keep in mind each model has different parameters 

      Common problems

      • Trademarked / Copyright
      • Political figures / Celebrities 
      • if it has children or child like images (an absolute no/ firefly will not generate anything that even resembles a child/ even childlike)

      or

      • - the prompt has words associated with children example: boy, girl, kid, kids, child, children and so on...
        Many Generation Models will not generate anything to do with children.
      • Nudity

      Cheers

      Nate

      varadatech
      varadatechAuthor
      Participant
      April 24, 2026

      Yes here is the prompt:
      “

      3D animated cinematic sequence in EXACT same Pixar-style visual language as reference image, identical composition, camera angle, lens (35mm equivalent), framing, character proportions, environment layout, and lighting setup, Rohan seated at wooden desk (right side of frame) writing in open notebook, anthropomorphic pencil character standing on desk (left side), warm indoor ambient lighting from right-side lamp, soft global illumination, subtle bounce light from desk surface, shallow depth of field focused on characters, background softly blurred but readable (educational posters, toys, shelves), perfectly consistent scene continuity, NO camera repositioning

      ABSOLUTE DIALOGUE SYNC SYSTEM (ZERO ERROR GUARANTEE):

      • Rohan voice MUST originate ONLY from Rohan character, perfectly lip-synced to his mouth movement

      • Pencil voice MUST originate ONLY from pencil character, perfectly synced to its facial deformation

      • NO voice overlap, NO off-screen dialogue, NO swapping voices under any condition

      • Each line triggered ONLY when that character is visibly speaking

      • Insert 0.3–0.5 second natural pauses between speaker changes to prevent overlap

      REALISTIC PERFORMANCE RULE:

      • All acting grounded in realism with subtle stylization

      • NO exaggerated cartoon explosions, NO over-the-top squash unless physically justified

      • Micro-expressions, breathing, blinking, eye focus shifts included

      • Natural timing and weight in all movements

      SCENE TIMELINE (LOCKED TIMING):

      [0:08 – 0:09.5]
      Medium shot (same as previous scene), Rohan pauses writing mid-stroke, pencil tip still touching paper, slight wrist freeze
      Rohan slowly turns head toward pencil, eyebrows lift, eyes narrow slightly in confusion
      Lip-sync ONLY on Rohan:
      “Wait… did YOU just sniff?”
      Pencil remains still during this line (NO movement, NO sound)

      [0:09.5 – 0:10.5]
      Pencil begins subtle pre-sneeze behavior:

      • tiny inhale motion (body slightly expands)

      • 2 very soft sniff actions (micro forward twitch, slight vibration)

      • minimal nose twitch deformation

      • very slight shiver (low amplitude)
        Lip-sync ONLY on pencil:
        “I think… I’m catching a cold 🤧”
        Voice slightly congested, soft delivery, no exaggeration

      [0:10.5 – 0:11.2]
      Rohan reacts:
      leans back ~5–8 cm, eyebrows furrow, slight disbelief expression
      hand holding pen lifts slightly off paper
      Lip-sync ONLY on Rohan:
      “Pencils don’t get sick!”
      Natural tone, not exaggerated

      [0:11.2 – 0:12]
      Pencil prepares sneeze (REALISTIC PHYSICS BUILD-UP):

      • small inhale (body expands subtly)

      • 0.2 sec pause (tension)
        Lip-sync ONLY on pencil:
        “WATCH ME… ACHOO!!”

      SNEEZE EXECUTION (PHYSICALLY ACCURATE — NO EXPLOSION):

      • short forward motion (2–3 cm displacement)

      • controlled air burst (not violent)

      • emission of fine graphite/ink micro-particles (NOT large blobs)

      • directional spread cone ~25–35 degrees

      • particle velocity moderate, quickly decaying

      FLUID + MATERIAL INTERACTION:

      • particles land on notebook surface

      • ink/graphite spreads via capillary action into paper fibers

      • creates organic, soft-edged scribbles and stains

      • absorption delay ~0.2–0.4 sec for realism

      • NO splash explosion, NO unrealistic coverage

      ROHAN REACTION:

      • quick flinch backward (head + shoulders)

      • blink reflex within 0.1 sec

      • slight recoil, then freeze

      • maintain believability (no exaggerated jump)

      ENVIRONMENT INTERACTION:

      • minimal particle spill onto desk (very subtle)

      • no environmental destruction

      • lighting remains stable and physically consistent

      ANIMATION QUALITY:

      • ultra-smooth animation curves

      • correct weight, inertia, and timing

      • no jitter, no clipping, no deformation errors

      • precise facial rigging and lip-sync accuracy

      • correct hand anatomy and contact with pen

      RENDERING:
      ultra high resolution, 8K, ultra detailed, sharp focus, crisp edges, professional lighting, realistic shadows, global illumination, ray-traced rendering, high dynamic range, photorealistic materials, accurate subsurface scattering, detailed wood grain, realistic paper texture, advanced particle simulation, cinematic depth of field, premium commercial quality, no artifacts, no distortion, no blur, no noise

      ”

      varadatech
      varadatechAuthor
      Participant
      April 24, 2026

      appreciate the clarification.

      What would really help on our end is having a more structured reference to follow across different models.

      Specifically, is there any official guide or documentation that breaks down:

      • Model-specific prompting rules
      • What should be explicitly included in prompts vs. what should be avoided
      • Any differences in how each model interprets structure, tags, or constraints

      Right now, it feels like we’re guessing per model, which makes consistency difficult—especially for more complex prompts like the one shared.

      If there’s a standardized framework or even product-level guidance we can rely on, that would make a big difference in aligning prompts correctly and avoiding silent failures like this.

      Would love to follow the intended approach rather than trial-and-error.

      Terms & ConditionsAccessibility statement
      Using the community Terms of use Privacy Cookie preferences Do not sell or share my personal information ad choicesAdChoices