Maintenance mode — queue paused
Queue
⚡ Cloud failover active
47 videos today
—
Generate
New video-to-video job
Who is making this?
H
Harry
O
Owen
F
Fenton
Select a user to continue
Reference video
Drop reference video or photo
MP4 · MOV · JPG · PNG · max 200 MB
First frame preview
Mode
Single
A/B — two outputs
Batch
Model
Ivy
Fashion · editorial · portrait
Hollie
Product · commercial · studio
Length
5
sec
1s10s20s30s
Output will match the length of your input video
Style
Voice · RVC
Ivy voice
Trained · consistent per-model
Hollie voice
Trained · consistent per-model
RVC runs automatically after generation. Reference audio is stripped and replaced. Voice is consistent across every video using this model — it never sounds like the original talent.
0
Video controls
50%
Low = closer to reference · High = more creative · Dancing sweet spot: 40–60%
7.0
80%
70%
75%
100%
Options
Face lock
Keep face consistent across all frames
Lip sync mode
Prioritise facial accuracy — for talking videos
Audio preserve
Carry original audio through to output
Background replace
Remove and replace the reference background
Auto-retry on failure
Requeue automatically if the job crashes
Pin a specific first frame
Generation starts from this exact image
Select a user before adding to queue
Queue
Running and scheduled jobs
3
In queue
~18m
Clears in
47
Done today
+6 vs yesterday
Harry Owen Fenton
Queue paused — current job finishes, no new jobs start
dance_ref_01 → Ivy · warm studio · denoising 50%
GPU 1 · 17 sec · standard · seed 4829301 · face lock · audio preserve
2× upscale2× RIFERVC · Ivy voice
✓Generate
✓Upscale
✓RIFE
◉RVC
○Merge
RVC voice conversion
GPU 3 · 3090 · replacing audio track
~35 sec remaining 13%
Running
talk_ref_03 → Hollie · cool editorial A/B · A
Queued · 5 sec · draft · seed 7723901 · lip sync on
Ready in ~11 min
Next
talk_ref_03 → Hollie · cool editorial A/B · B
Queued · 5 sec · draft · seed 3341820 · lip sync on
Ready in ~14 min
Pos 2
dance_ref_02 → Ivy · dramatic · seed 2201984
Done · 3 min ago · 30 sec · D:\harry\dance_ref_02_out.mp4
Done
Hardware
GPU 1 · 5060 TiGenerating
71°
Temp
VRAM 9.3 / 16 GB
Load 95%
GPU 2 · 5060 TiIdle
34°
Temp
VRAM 1.2 / 16 GB
Load 2%
GPU 3 · 3090Upscaling
62°
Temp
VRAM 11.4 / 24 GB
Load 88%
Storage · 2TB drive
847 GB used1,153 GB free
Harry — 312 GB
Owen — 284 GB
Fenton — 251 GB
Schedule
Batch jobs run overnight automatically
Jobs run at the set time. PC and ComfyUI must be on. GPU alert thresholds still apply.
New scheduled job
Who
H
Harry
O
Owen
F
Fenton
Drop reference videos
All processed with the same settings
Scheduled
dance_batch_10 → Ivy · dramatic
Tonight · 23:00 · 10 files · 30 sec · high · face lock
talk_weekly → Hollie · warm studio
Every Monday · 06:00 · 5 files · 5 sec · lip sync
editorial_overnight → Ivy
Fri 12 Sep · 22:30 · 8 files · 17 sec · high quality
Compare
Side by side — pick the winner
0:05
talk_ref_03 A/B·A
Hollie · Owen · seed 7723901 · denoising 50% · lip sync
0:05
talk_ref_03 A/B·B
Hollie · Owen · seed 3341820 · denoising 50% · lip sync
Output trimmer
0:000:17
In: 0:02Out: 0:15
Ref library
Pick from here instead of re-uploading every time
dance_fullbody_01
Dancing · 30s · full body
talk_closeup_01
Talking · 15s · close up
dance_waist_up
Dancing · 17s · waist up
talk_medium_02
Talking · 9s · medium shot
dance_slow_01
Dancing · 30s · slow motion
talk_side_profile
Talking · 12s · side angle
dance_energy_02
Dancing · 22s · full body
talk_direct_cam
Talking · 5s · direct to camera
Projects
Organise outputs into folders
September editorial
87 videos · Harry, Fenton · last active today
Active
Dance content batch Q3
312 videos · Owen · last active 2 days ago
Active
Talking head series — Aug
541 videos · all users · completed Aug 2026
Archive
Templates
Saved job settings — load in one click
Quick dance draft
Ivy · cinematic · 5 sec · draft
Denoising 50% · motion fidelity 80%
Face lock · warm studio
Full dance 30s high
Ivy · cinematic · 30 sec · high
Denoising 50% · 2× upscale · 2× RIFE
Face lock · audio preserve · dramatic
Talking — lip sync
Hollie · studio · match input · standard
Lip sync · face lock · audio preserve
Denoising 35% · CFG 7.5
Overnight batch high
Ivy · cinematic · match input · high
4× upscale · 4× RIFE · face lock
Denoising 55% · motion fidelity 85%
Health
System status · GPU control
ComfyUI
Online
Port 8188 · responding
Cloudflare
Connected
Tunnel healthy
Failover
Standby
DO API reachable
2TB drive
Healthy
847 GB used
GPU 1 temp
71°C
Alert at 80°C
GPU 2 temp
34°C
Alert at 80°C
GPU management
GPU 1 · RTX 5060 Ti
16 GB GDDR7 · Slot 1 · Generation
Online
Role
Generate
Upscale
RIFE
RVC
All post
Live stats
Temp
71°C
VRAM
9.3G
Load
95%
⚠ GPU 1 is offline — jobs will not route to this card until restarted
GPU 2 · RTX 5060 Ti
16 GB GDDR7 · Slot 2 · Generation
Idle
Role
Generate
Upscale
RIFE
RVC
All post
Live stats
Temp
34°C
VRAM
1.2G
Load
2%
⚠ GPU 2 is offline — jobs will not route to this card until restarted
GPU 3 · RTX 3090
24 GB GDDR6X · Slot 3 · Upscale + RIFE + RVC
Post
Role
Generate
Upscale
RIFE
RVC
All post
Live stats
Temp
62°C
VRAM
11.4G
Load
88%
⚠ GPU 3 is offline — post-processing paused until restarted
Alert thresholds
GPU 1 · 5060 Ti
80°C
Currently 71°C — 9° below threshold
GPU 2 · 5060 Ti
80°C
Currently 34°C — well below threshold
GPU 3 · 3090
85°C
Currently 62°C — 23° below threshold
Auto-retry and VRAM
Auto-retry on failure
Requeue job if GPU crashes — max 2 retries per job
GPU 1 — VRAM peak today
11.2 GB
of 16 GB
GPU 2 — VRAM peak today
3.4 GB
of 16 GB
GPU 3 — VRAM peak today
14.8 GB
of 24 GB
Analytics
Throughput, GPU hours, usage by user
47
Today
+6 vs yesterday
312
This week
1,204
All time
~£1.50
Today's electricity
Videos per hour — today
8am
9am
10am
11am
12pm
1pm
2pm
3pm
4pm
5pm
GPU hours
GPU 1 — today
6.4 hrs
of ~8hr working day
GPU 2 — today
1.2 hrs
mostly idle today
GPU 1 — this week
28.3 hrs
across 5 days
GPU 2 — this week
9.7 hrs
across 5 days
By user — this week
142
Harry
107
Owen
63
Fenton
Average generation time
5 sec · draft
2.1 min
17 sec · standard
7.4 min
30 sec · high
18.2 min
Train model
Upload references · 30–90 min · model appears in Generate once done
Drop 10–20 reference photos
JPG · PNG · consistent subject · varied angles
Train voice · RVC
Record or upload 10+ minutes of clean speech audio for the voice you want to use. RVC training takes ~15 minutes. One voice per visual model — Ivy has one voice, Hollie has one voice.
Drop voice audio
WAV · MP3 · min 10 minutes · clean speech · no music
Your voices
Ivy voice
Assigned to Ivy · trained 3 days ago · 14 min audio · pitch 0
Active
Hollie voice
Assigned to Hollie · trained 1 week ago · 22 min audio · pitch 0
Active
Your visual models
Ivy
Fashion · editorial · portrait · trained 3 days ago · 18 photos
Active
Hollie
Product · commercial · studio · trained 1 week ago · 22 photos
Active
Settings
ComfyUI endpoint, failover, users
ComfyUI
Failover
Users
Harry
D:\harry\
Owen
D:\owen\
Fenton
D:\fenton\