Mera 2.0
The flagship Omni model. Build a scene where motion, sound, and emotion stay connected for up to two minutes.
Mera 2.0
Every frame. Every feeling.
Mera 2.0 is a flagship video generation model by Skytells, bringing believable motion, realism, lip sync, audio, and emotional performance into the same scene.
Video. Audio. A performance that belongs together.
The Mera family
Mera is Omni.
Text to video. Image to video. Audio-driven lip sync, avatars, and reference-guided creation. Mera brings these modes into one Omni family. Mera 2.0 and Mera 2.0 Avatar can generate up to 120 seconds of video, maintaining a consistent subject and emotional continuity with a deep understanding of meaning.
Mera Avatar
The avatar family. Mera 2.0 Avatar carries a consistent subject and emotional performance through videos of up to 120 seconds.
Mera Recast
The Recast family extends Mera’s creative possibilities, alongside the flagship and Avatar models.
A demanding test. Every word, felt.
We chose Classical Arabic to challenge Mera’s pronunciation, phrasing, and emotional understanding. Its precise articulation and sustained vocal passages ask a great deal of even an experienced speaker. A word of love can carry longing, tenderness, or loss. Watch how the face follows that meaning as you listen to the voice.
Motion with weight.
Physics that gives a scene its grounding. Movement, contact, and momentum are designed to feel connected to the world around them.
Realism in the details.
Light, texture, and human expression work together. The small details make the whole scene feel more convincing.
A voice. A performance.
Audio and lip sync belong to the same moment. Speech timing, facial delivery, and sound carry the scene together.
The voice leads. The feeling follows.
More than pronunciation.
Classical Arabic brings precise articulation, breath, and musical phrasing into one performance. Listen for the held notes. Watch how the mouth stays with the sound.
A change in tone. A change in expression.
Tenderness, longing, and sorrow can live in the same phrase. The voice carries those shifts; the eyes and face make them visible.
A performance that holds together.
A sustained vowel asks more of a scene than a single spoken syllable. Vocal timing, facial movement, and the continuity of the shot need to stay connected.
Voice understanding
Mera hears the voice.
Mera listens to the original audio, understands its emotional meaning, and adapts the face, lip sync, and movement to the voice. Models that receive only a transcript work from written words, without the speaker’s original tone or delivery.
Mera hears the voice. Other models read the words.
Mera
Hears the original voice. Understands its emotion.
Pitch & rhythmPauses & emphasisEmotional tone
Meaning & emotion Expression & motion
Mera hears pitch, breath, pauses, and emotion in the original voice. It understands the meaning behind the delivery and adapts lip sync, facial expression, and movement to it.
Other models
Receive a transcript instead of hearing the voice.
Text input
The same phrase, written down.
Words & punctuation
Text interpretation Generated delivery
A transcript contains the words, but leaves out the sound of the original performance. These models infer emotion from text and generate delivery without hearing the speaker’s tone.
| What shapes the performance | Mera | Other models |
|---|---|---|
| Original voice | Heard directly | Transcribed into words |
| Emotional meaning | Understood through voice and delivery | Inferred from written text |
| Expression & lip sync | Adapt to the speaker’s vocal emotion | Generated from text instructions |
Audio understanding
It hears the voice. Not just the words.
Mera listens to the voice itself. The rhythm of a sentence. A hesitation. The lift in a laugh. Vocal delivery gives it context for emotion, expression, and lip sync.
A transcript tells you what was said.
A system that reduces speech to plain text keeps the words. Pitch, intensity, pauses, and the texture of the original voice can be left behind. Generating a new voice from that text does not automatically recover the original delivery.
The voice tells you how it was meant.
Listening to audio preserves the cues that give the same words a different feeling. Mera uses that vocal context to connect the sound of a performance with the face and movement on screen.
Why delivery matters: research on combining audio and transcripts for emotion recognition.
Get to know Mera 2.0.
Mera 2.0 is a flagship video generation model by Skytells. It brings motion, visual realism, audio, lip sync, and emotional performance into a single generated scene.
Mera uses the sound of the voice, including rhythm, pitch, pauses, and delivery, as context for a performance. A plain transcript preserves the words but can leave those vocal cues behind. The voice helps connect sound with expression and movement.
We chose Classical Arabic as a demanding test of articulation, sustained notes, and emotional delivery. Its formal register differs from everyday spoken Arabic. Watch the mouth, eyes, and expression through each phrase with sound on.
No. Mera is an Omni model supporting text-to-video, image-to-video, audio-driven lip sync, avatars, and reference-guided creation. We chose Classical Arabic as a demanding test for these previews. Its creative scope extends beyond this language and performance.
Mera 2.0 and Mera 2.0 Avatar can generate up to 120 seconds of video while maintaining a consistent subject and emotional continuity. The settings and models available to you depend on your workspace and plan.
The Mera family includes the flagship Omni model, Mera Avatar, and Mera Recast. This page showcases Mera 2.0; model availability in Relam is managed by your workspace.
Lip sync connects mouth movement with sound. Mera’s performance also brings facial expression and movement into the scene. In the preview, watch how the face follows changes in vocal delivery, rather than looking only at the lips.
Look at movement, contact, and momentum, then at lighting, texture, and continuity across frames. In the vocal example, sustained notes make it easier to inspect whether facial movement stays connected to the sound.
Open Relam and describe the scene you want to create in chat. Available models, generation settings, and access depend on your workspace and plan. You can also upload an image to edit or use as a reference when the selected model supports it.