refwarm100k ep018 / ep019 vs mainline ep018 (+ InstructAV2AV) — 100 test clips

test_fpsfix_rand100_seed0 (test_fpsfix_rand50_seed0 + 50 random). Removal without the first-frame anchor, addition with it; same seed per clip for both models. Video vs each cell's GT target.mp4; audio IS/KL from the cnn14 classifier (audioldm_eval), FAD on the login CPU.

cell · model	PSNR↑	SSIM↑	LPIPS↓	FVD↓	FAD↓	IS↑	KL↓	LSD↓
removal (no first frame) · mainline ep018	20.17	0.657	0.291	614	2.44	3.28	1.132	14.39
removal (no first frame) · refwarm100k ep018	19.59	0.647	0.308	634	2.42	3.32	1.162	14.55
removal (no first frame) · refwarm100k ep019	19.78	0.642	0.300	707	2.49	3.32	1.217	14.34
addition (first frame) · mainline ep018	16.49	0.556	0.323	1185	3.54	3.31	1.108	12.11
addition (first frame) · refwarm100k ep018	19.08	0.623	0.230	765	3.39	3.29	1.194	12.03
addition (first frame) · refwarm100k ep019	19.15	0.622	0.229	858	3.54	3.40	1.176	12.10

Paired per clip, refwarm ep018 − mainline ep018 (refwarm better / worse, Wilcoxon p). FVD/FAD are set-level and noisy at n=100.

pair	ΔPSNR	ΔSSIM	ΔLPIPS	ΔKL	ΔLSD
removal (no first frame): refwarm100k ep019 − refwarm100k ep018	+0.185 (53/47, p=0.42)	-0.004 (41/59, p=0.23)	-0.008 (47/53, p=0.57)	+0.056 (48/52, p=0.95)	-0.208 (47/53, p=0.9)
removal (no first frame): refwarm100k ep019 − mainline ep018	-0.393 (39/61, p=0.02)	-0.015 (33/67, p=0.00043)	+0.009 (42/58, p=0.041)	+0.085 (50/50, p=0.94)	-0.044 (48/52, p=0.81)
addition (first frame): refwarm100k ep019 − refwarm100k ep018	+0.074 (48/52, p=0.5)	-0.001 (30/70, p=0.00039)	-0.001 (37/63, p=0.22)	-0.018 (41/59, p=0.51)	+0.076 (48/52, p=0.38)
addition (first frame): refwarm100k ep019 − mainline ep018	+2.656 (91/9, p=4.2e-16)	+0.066 (85/15, p=5.9e-13)	-0.094 (94/6, p=1.5e-17)	+0.068 (44/56, p=0.16)	-0.003 (56/44, p=0.99)

Verdict (100 test clips). Addition (with first frame): refwarm ep018 is much better: +2.58 dB PSNR (93/100 clips better), SSIM +0.067 (87/100), LPIPS −0.093 (94/100), all p<1e-10; FVD 765 vs 1185. Audio is a tie (KL, LSD not significant; FAD 3.39 vs 3.54). Removal (no first frame): mainline ep018 is better, by a smaller margin: PSNR +0.58 dB (63/100 clips, p=0.002), SSIM (p=0.02), LPIPS (p=0.001); FVD 614 vs 634; audio tie. So refwarm's in-distribution anchor fixes addition, at a small cost in removal.

All 100 test examples

Every clip of test_fpsfix_rand100_seed0 (random, not cherry-picked). Captions: per-clip PSNR / LPIPS / LSD vs the GT target. InstructAV2AV (last column) is shown for visual comparison only; it is not in the tables above.

IAlYlq8L4ug_000010 — “wind instrument”

removal (no first frame)

input

mainline ep018 · PSNR 24.31 · LPIPS 0.099 · LSD 10.84

refwarm100k ep018 · PSNR 22.93 · LPIPS 0.144 · LSD 10.81

refwarm100k ep019 · PSNR 22.99 · LPIPS 0.146 · LSD 10.97

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 22.52 · LPIPS 0.128 · LSD 11.17

refwarm100k ep018 · PSNR 23.83 · LPIPS 0.082 · LSD 11.05

refwarm100k ep019 · PSNR 22.93 · LPIPS 0.090 · LSD 11.54

InstructAV2AV (first frame)

XD059jkt6bs — “bowed string instrument”

removal (no first frame)

input

mainline ep018 · PSNR 25.31 · LPIPS 0.193 · LSD 15.43

refwarm100k ep018 · PSNR 25.33 · LPIPS 0.195 · LSD 15.50

refwarm100k ep019 · PSNR 23.89 · LPIPS 0.224 · LSD 15.40

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 14.46 · LPIPS 0.412 · LSD 13.97

refwarm100k ep018 · PSNR 17.33 · LPIPS 0.351 · LSD 13.73

refwarm100k ep019 · PSNR 17.42 · LPIPS 0.364 · LSD 13.83

InstructAV2AV (first frame)

wSf_whw3c5A — “keyboard-like instrument”

removal (no first frame)

input

mainline ep018 · PSNR 21.24 · LPIPS 0.205 · LSD 14.61

refwarm100k ep018 · PSNR 17.99 · LPIPS 0.349 · LSD 14.78

refwarm100k ep019 · PSNR 16.29 · LPIPS 0.272 · LSD 14.51

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 16.91 · LPIPS 0.216 · LSD 12.49

refwarm100k ep018 · PSNR 18.38 · LPIPS 0.164 · LSD 12.37

refwarm100k ep019 · PSNR 18.57 · LPIPS 0.162 · LSD 11.89

InstructAV2AV (first frame)

DlulAwmcYcE — “a person”

removal (no first frame)

input

mainline ep018 · PSNR 24.21 · LPIPS 0.154 · LSD 10.53

refwarm100k ep018 · PSNR 14.54 · LPIPS 0.457 · LSD 10.46

refwarm100k ep019 · PSNR 14.98 · LPIPS 0.283 · LSD 10.16

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 24.44 · LPIPS 0.118 · LSD 10.69

refwarm100k ep018 · PSNR 19.11 · LPIPS 0.217 · LSD 10.65

refwarm100k ep019 · PSNR 22.50 · LPIPS 0.143 · LSD 10.66

InstructAV2AV (first frame)

PpPEu1wNM4E — “engine”

removal (no first frame)

input

mainline ep018 · PSNR 14.89 · LPIPS 0.458 · LSD 10.42

refwarm100k ep018 · PSNR 14.75 · LPIPS 0.550 · LSD 10.46

refwarm100k ep019 · PSNR 14.54 · LPIPS 0.582 · LSD 10.18

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 13.42 · LPIPS 0.476 · LSD 9.63

refwarm100k ep018 · PSNR 17.75 · LPIPS 0.275 · LSD 9.69

refwarm100k ep019 · PSNR 17.93 · LPIPS 0.280 · LSD 9.63

InstructAV2AV (first frame)

SDCdqZuGMqc_000035 — “cat”

removal (no first frame)

input

mainline ep018 · PSNR 16.03 · LPIPS 0.334 · LSD 34.41

refwarm100k ep018 · PSNR 15.57 · LPIPS 0.333 · LSD 33.84

refwarm100k ep019 · PSNR 14.48 · LPIPS 0.421 · LSD 33.99

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 14.94 · LPIPS 0.406 · LSD 12.93

refwarm100k ep018 · PSNR 21.06 · LPIPS 0.188 · LSD 14.40

refwarm100k ep019 · PSNR 20.98 · LPIPS 0.192 · LSD 12.40

InstructAV2AV (first frame)

aOR4wN_vTv8_000030 — “water”

removal (no first frame)

input

mainline ep018 · PSNR 31.75 · LPIPS 0.060 · LSD 14.32

refwarm100k ep018 · PSNR 30.58 · LPIPS 0.065 · LSD 13.95

refwarm100k ep019 · PSNR 31.08 · LPIPS 0.059 · LSD 14.52

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 23.33 · LPIPS 0.179 · LSD 12.80

refwarm100k ep018 · PSNR 29.52 · LPIPS 0.072 · LSD 12.79

refwarm100k ep019 · PSNR 28.50 · LPIPS 0.081 · LSD 12.29

InstructAV2AV (first frame)

ttANmdLLQN4_000022 — “guitar”

removal (no first frame)

input

mainline ep018 · PSNR 20.96 · LPIPS 0.147 · LSD 11.02

refwarm100k ep018 · PSNR 19.96 · LPIPS 0.195 · LSD 11.17

refwarm100k ep019 · PSNR 19.80 · LPIPS 0.205 · LSD 11.05

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.68 · LPIPS 0.258 · LSD 10.27

refwarm100k ep018 · PSNR 22.38 · LPIPS 0.095 · LSD 10.37

refwarm100k ep019 · PSNR 21.81 · LPIPS 0.109 · LSD 10.26

InstructAV2AV (first frame)

FoYY0o6JzYI — “engine”

removal (no first frame)

input

mainline ep018 · PSNR 14.06 · LPIPS 0.516 · LSD 10.08

refwarm100k ep018 · PSNR 13.44 · LPIPS 0.575 · LSD 9.97

refwarm100k ep019 · PSNR 13.71 · LPIPS 0.564 · LSD 9.61

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 12.22 · LPIPS 0.589 · LSD 9.36

refwarm100k ep018 · PSNR 14.45 · LPIPS 0.507 · LSD 9.29

refwarm100k ep019 · PSNR 14.35 · LPIPS 0.509 · LSD 9.18

InstructAV2AV (first frame)

zIFanxFCPmo — “engine”

removal (no first frame)

input

mainline ep018 · PSNR 14.28 · LPIPS 0.431 · LSD 14.35

refwarm100k ep018 · PSNR 15.93 · LPIPS 0.440 · LSD 14.19

refwarm100k ep019 · PSNR 13.78 · LPIPS 0.602 · LSD 12.91

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 8.81 · LPIPS 0.556 · LSD 9.58

refwarm100k ep018 · PSNR 11.50 · LPIPS 0.384 · LSD 9.95

refwarm100k ep019 · PSNR 12.73 · LPIPS 0.312 · LSD 9.73

InstructAV2AV (first frame)

1gw5xoQef_s_000025 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 18.83 · LPIPS 0.353 · LSD 15.56

refwarm100k ep018 · PSNR 19.30 · LPIPS 0.311 · LSD 19.04

refwarm100k ep019 · PSNR 18.69 · LPIPS 0.288 · LSD 19.57

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 12.22 · LPIPS 0.512 · LSD 18.78

refwarm100k ep018 · PSNR 15.62 · LPIPS 0.352 · LSD 19.01

refwarm100k ep019 · PSNR 14.99 · LPIPS 0.362 · LSD 15.31

InstructAV2AV (first frame)

7dK7_x9tYFs — “sewing machine”

removal (no first frame)

input

mainline ep018 · PSNR 16.27 · LPIPS 0.521 · LSD 12.72

refwarm100k ep018 · PSNR 17.40 · LPIPS 0.421 · LSD 13.06

refwarm100k ep019 · PSNR 17.62 · LPIPS 0.398 · LSD 13.24

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 21.51 · LPIPS 0.162 · LSD 10.79

refwarm100k ep018 · PSNR 21.80 · LPIPS 0.145 · LSD 10.62

refwarm100k ep019 · PSNR 21.11 · LPIPS 0.163 · LSD 10.70

InstructAV2AV (first frame)

GQc2wwoTCQ4_000070 — “horse”

removal (no first frame)

input

mainline ep018 · PSNR 22.68 · LPIPS 0.211 · LSD 11.94

refwarm100k ep018 · PSNR 22.13 · LPIPS 0.225 · LSD 11.43

refwarm100k ep019 · PSNR 22.36 · LPIPS 0.224 · LSD 11.42

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 19.41 · LPIPS 0.273 · LSD 11.26

refwarm100k ep018 · PSNR 20.65 · LPIPS 0.250 · LSD 11.53

refwarm100k ep019 · PSNR 20.75 · LPIPS 0.253 · LSD 11.33

InstructAV2AV (first frame)

0WG4M3HAgv4_000331 — “food”

removal (no first frame)

input

mainline ep018 · PSNR 11.45 · LPIPS 0.494 · LSD 13.48

refwarm100k ep018 · PSNR 11.77 · LPIPS 0.550 · LSD 13.74

refwarm100k ep019 · PSNR 13.96 · LPIPS 0.428 · LSD 14.51

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 13.54 · LPIPS 0.309 · LSD 17.03

refwarm100k ep018 · PSNR 19.64 · LPIPS 0.109 · LSD 17.29

refwarm100k ep019 · PSNR 20.09 · LPIPS 0.102 · LSD 17.47

InstructAV2AV (first frame)

uAOE4m9xNt0_000079 — “water”

removal (no first frame)

input

mainline ep018 · PSNR 14.92 · LPIPS 0.764 · LSD 9.91

refwarm100k ep018 · PSNR 14.78 · LPIPS 0.745 · LSD 9.81

refwarm100k ep019 · PSNR 16.66 · LPIPS 0.561 · LSD 10.09

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 12.17 · LPIPS 0.637 · LSD 9.09

refwarm100k ep018 · PSNR 12.55 · LPIPS 0.573 · LSD 9.04

refwarm100k ep019 · PSNR 14.43 · LPIPS 0.464 · LSD 9.02

InstructAV2AV (first frame)

65vCgp9aKGU_000116 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 18.86 · LPIPS 0.651 · LSD 15.48

refwarm100k ep018 · PSNR 18.74 · LPIPS 0.648 · LSD 15.33

refwarm100k ep019 · PSNR 19.12 · LPIPS 0.633 · LSD 18.04

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 13.46 · LPIPS 0.592 · LSD 14.12

refwarm100k ep018 · PSNR 14.41 · LPIPS 0.541 · LSD 13.03

refwarm100k ep019 · PSNR 14.41 · LPIPS 0.551 · LSD 17.93

InstructAV2AV (first frame)

iXAr-dqiWxA_000090 — “large vehicle”

removal (no first frame)

input

mainline ep018 · PSNR 23.11 · LPIPS 0.161 · LSD 14.13

refwarm100k ep018 · PSNR 23.14 · LPIPS 0.153 · LSD 14.49

refwarm100k ep019 · PSNR 23.14 · LPIPS 0.151 · LSD 14.06

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 19.23 · LPIPS 0.224 · LSD 10.18

refwarm100k ep018 · PSNR 19.71 · LPIPS 0.207 · LSD 10.05

refwarm100k ep019 · PSNR 19.57 · LPIPS 0.209 · LSD 10.96

InstructAV2AV (first frame)

8iyZSaEWYNE — “vehicle engine”

removal (no first frame)

input

mainline ep018 · PSNR 24.30 · LPIPS 0.094 · LSD 24.39

refwarm100k ep018 · PSNR 19.31 · LPIPS 0.297 · LSD 24.86

refwarm100k ep019 · PSNR 23.86 · LPIPS 0.102 · LSD 23.62

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 22.46 · LPIPS 0.110 · LSD 12.69

refwarm100k ep018 · PSNR 23.23 · LPIPS 0.082 · LSD 13.21

refwarm100k ep019 · PSNR 23.19 · LPIPS 0.083 · LSD 12.68

InstructAV2AV (first frame)

VV0ge1x5Jrw — “blender”

removal (no first frame)

input

mainline ep018 · PSNR 18.63 · LPIPS 0.324 · LSD 12.03

refwarm100k ep018 · PSNR 18.73 · LPIPS 0.318 · LSD 10.08

refwarm100k ep019 · PSNR 19.57 · LPIPS 0.272 · LSD 10.31

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 16.43 · LPIPS 0.360 · LSD 10.43

refwarm100k ep018 · PSNR 17.13 · LPIPS 0.355 · LSD 10.50

refwarm100k ep019 · PSNR 16.94 · LPIPS 0.351 · LSD 10.25

InstructAV2AV (first frame)

jsagp9KB-rk_000025 — “percussion piece”

removal (no first frame)

input

mainline ep018 · PSNR 13.81 · LPIPS 0.551 · LSD 11.69

refwarm100k ep018 · PSNR 16.21 · LPIPS 0.365 · LSD 12.52

refwarm100k ep019 · PSNR 16.79 · LPIPS 0.336 · LSD 12.51

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.58 · LPIPS 0.351 · LSD 12.44

refwarm100k ep018 · PSNR 19.03 · LPIPS 0.144 · LSD 12.46

refwarm100k ep019 · PSNR 19.27 · LPIPS 0.142 · LSD 14.57

InstructAV2AV (first frame)

H71KwdRWJkg_000292 — “sitar”

removal (no first frame)

input

mainline ep018 · PSNR 15.84 · LPIPS 0.336 · LSD 12.24

refwarm100k ep018 · PSNR 14.04 · LPIPS 0.466 · LSD 11.67

refwarm100k ep019 · PSNR 16.07 · LPIPS 0.332 · LSD 11.36

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 10.13 · LPIPS 0.432 · LSD 11.17

refwarm100k ep018 · PSNR 9.65 · LPIPS 0.451 · LSD 11.12

refwarm100k ep019 · PSNR 9.74 · LPIPS 0.445 · LSD 11.07

InstructAV2AV (first frame)

3exeRhxU-8c_000090 — “harp”

removal (no first frame)

input

mainline ep018 · PSNR 23.25 · LPIPS 0.268 · LSD 11.83

refwarm100k ep018 · PSNR 23.67 · LPIPS 0.297 · LSD 11.68

refwarm100k ep019 · PSNR 22.95 · LPIPS 0.313 · LSD 11.69

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.94 · LPIPS 0.421 · LSD 11.20

refwarm100k ep018 · PSNR 18.14 · LPIPS 0.356 · LSD 10.99

refwarm100k ep019 · PSNR 18.23 · LPIPS 0.371 · LSD 11.09

InstructAV2AV (first frame)

GqmiOpwIl4U_000030 — “individual”

removal (no first frame)

input

mainline ep018 · PSNR 23.81 · LPIPS 0.060 · LSD 9.56

refwarm100k ep018 · PSNR 22.65 · LPIPS 0.067 · LSD 9.64

refwarm100k ep019 · PSNR 13.38 · LPIPS 0.422 · LSD 10.03

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 13.94 · LPIPS 0.219 · LSD 9.63

refwarm100k ep018 · PSNR 17.79 · LPIPS 0.129 · LSD 9.48

refwarm100k ep019 · PSNR 18.56 · LPIPS 0.117 · LSD 9.65

InstructAV2AV (first frame)

3NWTOn8uhy4_000030 — “water”

removal (no first frame)

input

mainline ep018 · PSNR 27.79 · LPIPS 0.142 · LSD 9.73

refwarm100k ep018 · PSNR 26.06 · LPIPS 0.140 · LSD 9.86

refwarm100k ep019 · PSNR 26.38 · LPIPS 0.143 · LSD 9.07

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 26.48 · LPIPS 0.188 · LSD 9.97

refwarm100k ep018 · PSNR 26.16 · LPIPS 0.161 · LSD 9.58

refwarm100k ep019 · PSNR 26.65 · LPIPS 0.162 · LSD 10.09

InstructAV2AV (first frame)

H-CFunrRvAw — “train”

removal (no first frame)

input

mainline ep018 · PSNR 22.01 · LPIPS 0.207 · LSD 9.88

refwarm100k ep018 · PSNR 21.63 · LPIPS 0.222 · LSD 11.25

refwarm100k ep019 · PSNR 21.41 · LPIPS 0.227 · LSD 10.50

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 18.58 · LPIPS 0.264 · LSD 9.93

refwarm100k ep018 · PSNR 20.57 · LPIPS 0.167 · LSD 9.72

refwarm100k ep019 · PSNR 20.43 · LPIPS 0.179 · LSD 10.21

InstructAV2AV (first frame)

vETYTWWpcfk — “canine”

removal (no first frame)

input

mainline ep018 · PSNR 18.47 · LPIPS 0.221 · LSD 22.61

refwarm100k ep018 · PSNR 18.75 · LPIPS 0.218 · LSD 24.63

refwarm100k ep019 · PSNR 23.32 · LPIPS 0.100 · LSD 23.79

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 20.81 · LPIPS 0.126 · LSD 10.57

refwarm100k ep018 · PSNR 24.14 · LPIPS 0.063 · LSD 10.54

refwarm100k ep019 · PSNR 24.46 · LPIPS 0.061 · LSD 9.99

InstructAV2AV (first frame)

Fkn-poTw_gA — “individual”

removal (no first frame)

input

mainline ep018 · PSNR 29.39 · LPIPS 0.111 · LSD 16.35

refwarm100k ep018 · PSNR 27.03 · LPIPS 0.142 · LSD 13.01

refwarm100k ep019 · PSNR 26.15 · LPIPS 0.164 · LSD 12.67

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 18.15 · LPIPS 0.277 · LSD 12.63

refwarm100k ep018 · PSNR 25.15 · LPIPS 0.160 · LSD 12.99

refwarm100k ep019 · PSNR 25.00 · LPIPS 0.167 · LSD 12.53

InstructAV2AV (first frame)

rd-TgpH5ds8_000030 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 15.88 · LPIPS 0.403 · LSD 15.56

refwarm100k ep018 · PSNR 15.57 · LPIPS 0.442 · LSD 13.80

refwarm100k ep019 · PSNR 15.52 · LPIPS 0.460 · LSD 14.97

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 9.21 · LPIPS 0.589 · LSD 9.96

refwarm100k ep018 · PSNR 9.38 · LPIPS 0.562 · LSD 10.19

refwarm100k ep019 · PSNR 9.29 · LPIPS 0.552 · LSD 11.04

InstructAV2AV (first frame)

X1U3soaFDxo — “car engine”

removal (no first frame)

input

mainline ep018 · PSNR 16.28 · LPIPS 0.393 · LSD 18.28

refwarm100k ep018 · PSNR 15.66 · LPIPS 0.420 · LSD 18.48

refwarm100k ep019 · PSNR 15.44 · LPIPS 0.438 · LSD 19.15

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 11.66 · LPIPS 0.449 · LSD 9.72

refwarm100k ep018 · PSNR 21.10 · LPIPS 0.120 · LSD 9.88

refwarm100k ep019 · PSNR 21.68 · LPIPS 0.101 · LSD 9.72

InstructAV2AV (first frame)

mkDIrir-tkY_000431 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 17.28 · LPIPS 0.323 · LSD 18.51

refwarm100k ep018 · PSNR 15.12 · LPIPS 0.433 · LSD 19.16

refwarm100k ep019 · PSNR 19.81 · LPIPS 0.251 · LSD 18.51

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 11.06 · LPIPS 0.401 · LSD 15.00

refwarm100k ep018 · PSNR 14.00 · LPIPS 0.287 · LSD 15.20

refwarm100k ep019 · PSNR 13.81 · LPIPS 0.294 · LSD 14.01

InstructAV2AV (first frame)

wwRKltt-mXo_000106 — “rooster”

removal (no first frame)

input

mainline ep018 · PSNR 18.23 · LPIPS 0.391 · LSD 34.27

refwarm100k ep018 · PSNR 18.18 · LPIPS 0.386 · LSD 27.50

refwarm100k ep019 · PSNR 18.21 · LPIPS 0.367 · LSD 29.41

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 13.62 · LPIPS 0.467 · LSD 17.57

refwarm100k ep018 · PSNR 14.69 · LPIPS 0.449 · LSD 17.96

refwarm100k ep019 · PSNR 14.65 · LPIPS 0.446 · LSD 18.09

InstructAV2AV (first frame)

BEcZaiqph4Q — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 23.32 · LPIPS 0.077 · LSD 14.16

refwarm100k ep018 · PSNR 23.51 · LPIPS 0.067 · LSD 14.33

refwarm100k ep019 · PSNR 23.52 · LPIPS 0.075 · LSD 14.35

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 22.35 · LPIPS 0.076 · LSD 15.20

refwarm100k ep018 · PSNR 23.61 · LPIPS 0.064 · LSD 15.23

refwarm100k ep019 · PSNR 23.40 · LPIPS 0.065 · LSD 15.34

InstructAV2AV (first frame)

hZdfoAdmCEI — “sewing machine”

removal (no first frame)

input

mainline ep018 · PSNR 16.11 · LPIPS 0.603 · LSD 9.20

refwarm100k ep018 · PSNR 17.61 · LPIPS 0.613 · LSD 9.16

refwarm100k ep019 · PSNR 16.22 · LPIPS 0.619 · LSD 9.23

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 12.93 · LPIPS 0.499 · LSD 10.41

refwarm100k ep018 · PSNR 20.83 · LPIPS 0.147 · LSD 11.06

refwarm100k ep019 · PSNR 21.62 · LPIPS 0.133 · LSD 10.94

InstructAV2AV (first frame)

Yye5rc1HOmI — “synth”

removal (no first frame)

input

mainline ep018 · PSNR 16.21 · LPIPS 0.320 · LSD 12.88

refwarm100k ep018 · PSNR 12.71 · LPIPS 0.323 · LSD 12.64

refwarm100k ep019 · PSNR 13.01 · LPIPS 0.344 · LSD 11.65

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 12.84 · LPIPS 0.304 · LSD 10.97

refwarm100k ep018 · PSNR 17.92 · LPIPS 0.157 · LSD 11.17

refwarm100k ep019 · PSNR 17.92 · LPIPS 0.158 · LSD 11.27

InstructAV2AV (first frame)

JevO8TnLhUM_000020 — “electric guitar”

removal (no first frame)

input

mainline ep018 · PSNR 22.57 · LPIPS 0.183 · LSD 9.89

refwarm100k ep018 · PSNR 20.54 · LPIPS 0.219 · LSD 10.41

refwarm100k ep019 · PSNR 21.01 · LPIPS 0.229 · LSD 10.98

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 14.95 · LPIPS 0.343 · LSD 10.90

refwarm100k ep018 · PSNR 17.80 · LPIPS 0.226 · LSD 10.16

refwarm100k ep019 · PSNR 17.38 · LPIPS 0.244 · LSD 10.40

InstructAV2AV (first frame)

kv4EF-cJ2Fc_000070 — “male voice”

removal (no first frame)

input

mainline ep018 · PSNR 12.02 · LPIPS 0.463 · LSD 10.12

refwarm100k ep018 · PSNR 12.03 · LPIPS 0.478 · LSD 10.18

refwarm100k ep019 · PSNR 12.21 · LPIPS 0.427 · LSD 10.15

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 12.61 · LPIPS 0.396 · LSD 12.85

refwarm100k ep018 · PSNR 20.38 · LPIPS 0.119 · LSD 11.50

refwarm100k ep019 · PSNR 20.32 · LPIPS 0.130 · LSD 12.46

InstructAV2AV (first frame)

K31lEXM8fVQ — “motor vehicle engine”

removal (no first frame)

input

mainline ep018 · PSNR 24.89 · LPIPS 0.072 · LSD 31.32

refwarm100k ep018 · PSNR 24.96 · LPIPS 0.074 · LSD 31.78

refwarm100k ep019 · PSNR 24.92 · LPIPS 0.071 · LSD 32.03

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.28 · LPIPS 0.247 · LSD 10.35

refwarm100k ep018 · PSNR 23.21 · LPIPS 0.096 · LSD 9.69

refwarm100k ep019 · PSNR 22.29 · LPIPS 0.090 · LSD 9.91

InstructAV2AV (first frame)

Xp-KcbwtzwM — “vehicle engine”

removal (no first frame)

input

mainline ep018 · PSNR 19.45 · LPIPS 0.294 · LSD 9.75

refwarm100k ep018 · PSNR 20.75 · LPIPS 0.295 · LSD 9.76

refwarm100k ep019 · PSNR 18.53 · LPIPS 0.321 · LSD 9.70

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.31 · LPIPS 0.556 · LSD 9.67

refwarm100k ep018 · PSNR 23.49 · LPIPS 0.284 · LSD 9.62

refwarm100k ep019 · PSNR 25.42 · LPIPS 0.202 · LSD 9.51

InstructAV2AV (first frame)

H3Ll_xbB20Q — “man playing cello”

removal (no first frame)

input

mainline ep018 · PSNR 21.67 · LPIPS 0.189 · LSD 10.36

refwarm100k ep018 · PSNR 21.31 · LPIPS 0.182 · LSD 10.29

refwarm100k ep019 · PSNR 21.44 · LPIPS 0.188 · LSD 10.29

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 24.62 · LPIPS 0.101 · LSD 10.59

refwarm100k ep018 · PSNR 25.85 · LPIPS 0.069 · LSD 10.29

refwarm100k ep019 · PSNR 25.50 · LPIPS 0.071 · LSD 11.92

InstructAV2AV (first frame)

ZRbAhRimdGU_000000 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 19.62 · LPIPS 0.285 · LSD 24.94

refwarm100k ep018 · PSNR 19.24 · LPIPS 0.313 · LSD 25.41

refwarm100k ep019 · PSNR 19.44 · LPIPS 0.316 · LSD 24.52

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 16.48 · LPIPS 0.384 · LSD 16.14

refwarm100k ep018 · PSNR 16.60 · LPIPS 0.338 · LSD 16.52

refwarm100k ep019 · PSNR 16.39 · LPIPS 0.333 · LSD 15.58

InstructAV2AV (first frame)

ua2CjciXjHQ — “train signal”

removal (no first frame)

input

mainline ep018 · PSNR 11.94 · LPIPS 0.552 · LSD 18.06

refwarm100k ep018 · PSNR 17.80 · LPIPS 0.342 · LSD 21.18

refwarm100k ep019 · PSNR 20.00 · LPIPS 0.250 · LSD 17.93

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 11.69 · LPIPS 0.521 · LSD 14.16

refwarm100k ep018 · PSNR 12.84 · LPIPS 0.359 · LSD 15.33

refwarm100k ep019 · PSNR 13.23 · LPIPS 0.333 · LSD 15.73

InstructAV2AV (first frame)

RT0BMLsUln0 — “engine”

removal (no first frame)

input

mainline ep018 · PSNR 25.09 · LPIPS 0.161 · LSD 9.81

refwarm100k ep018 · PSNR 24.20 · LPIPS 0.164 · LSD 9.99

refwarm100k ep019 · PSNR 24.96 · LPIPS 0.150 · LSD 9.96

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 20.60 · LPIPS 0.226 · LSD 9.16

refwarm100k ep018 · PSNR 21.84 · LPIPS 0.197 · LSD 9.22

refwarm100k ep019 · PSNR 21.96 · LPIPS 0.192 · LSD 9.47

InstructAV2AV (first frame)

LBGeXwh5zL8 — “idling engine”

removal (no first frame)

input

mainline ep018 · PSNR 18.77 · LPIPS 0.203 · LSD 12.09

refwarm100k ep018 · PSNR 18.79 · LPIPS 0.197 · LSD 13.41

refwarm100k ep019 · PSNR 18.62 · LPIPS 0.199 · LSD 15.24

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 19.71 · LPIPS 0.153 · LSD 11.16

refwarm100k ep018 · PSNR 23.97 · LPIPS 0.085 · LSD 10.73

refwarm100k ep019 · PSNR 23.71 · LPIPS 0.087 · LSD 11.87

InstructAV2AV (first frame)

83Trbg07opQ — “engine”

removal (no first frame)

input

mainline ep018 · PSNR 17.74 · LPIPS 0.571 · LSD 9.34

refwarm100k ep018 · PSNR 16.68 · LPIPS 0.597 · LSD 9.35

refwarm100k ep019 · PSNR 15.63 · LPIPS 0.624 · LSD 9.14

InstructAV2AV: no output for this clip

addition (first frame)

input

mainline ep018 · PSNR 10.11 · LPIPS 0.581 · LSD 9.61

refwarm100k ep018 · PSNR 10.47 · LPIPS 0.578 · LSD 9.22

refwarm100k ep019 · PSNR 10.81 · LPIPS 0.554 · LSD 9.67

InstructAV2AV (first frame)

avd-_a-D7Ck_000003 — “harp”

removal (no first frame)

input

mainline ep018 · PSNR 12.93 · LPIPS 0.277 · LSD 9.98

refwarm100k ep018 · PSNR 10.81 · LPIPS 0.328 · LSD 9.87

refwarm100k ep019 · PSNR 13.79 · LPIPS 0.263 · LSD 9.93

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 16.25 · LPIPS 0.177 · LSD 10.13

refwarm100k ep018 · PSNR 20.07 · LPIPS 0.103 · LSD 10.31

refwarm100k ep019 · PSNR 19.88 · LPIPS 0.106 · LSD 10.63

InstructAV2AV (first frame)

EALyxfU4qyM_000016 — “individuals”

removal (no first frame)

input

mainline ep018 · PSNR 30.91 · LPIPS 0.074 · LSD 11.38

refwarm100k ep018 · PSNR 27.96 · LPIPS 0.073 · LSD 11.15

refwarm100k ep019 · PSNR 29.09 · LPIPS 0.082 · LSD 11.48

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.10 · LPIPS 0.346 · LSD 11.09

refwarm100k ep018 · PSNR 18.34 · LPIPS 0.245 · LSD 10.73

refwarm100k ep019 · PSNR 18.42 · LPIPS 0.246 · LSD 14.41

InstructAV2AV (first frame)

RHLZXT_DRkQ — “stringed instruments”

removal (no first frame)

input

mainline ep018 · PSNR 12.88 · LPIPS 0.425 · LSD 13.44

refwarm100k ep018 · PSNR 12.98 · LPIPS 0.408 · LSD 13.40

refwarm100k ep019 · PSNR 12.66 · LPIPS 0.489 · LSD 13.12

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 11.96 · LPIPS 0.417 · LSD 13.61

refwarm100k ep018 · PSNR 14.11 · LPIPS 0.273 · LSD 13.47

refwarm100k ep019 · PSNR 14.65 · LPIPS 0.258 · LSD 13.34

InstructAV2AV (first frame)

VEphB6rwU_A_000002 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 22.05 · LPIPS 0.356 · LSD 18.20

refwarm100k ep018 · PSNR 20.98 · LPIPS 0.409 · LSD 18.05

refwarm100k ep019 · PSNR 22.06 · LPIPS 0.347 · LSD 19.38

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.49 · LPIPS 0.321 · LSD 17.51

refwarm100k ep018 · PSNR 19.88 · LPIPS 0.188 · LSD 17.28

refwarm100k ep019 · PSNR 20.09 · LPIPS 0.206 · LSD 12.75

InstructAV2AV (first frame)

vRA2Ux-xyKQ — “vehicle engine”

removal (no first frame)

input

mainline ep018 · PSNR 24.14 · LPIPS 0.106 · LSD 11.33

refwarm100k ep018 · PSNR 17.85 · LPIPS 0.136 · LSD 10.87

refwarm100k ep019 · PSNR 24.39 · LPIPS 0.109 · LSD 11.29

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.77 · LPIPS 0.335 · LSD 9.86

refwarm100k ep018 · PSNR 20.02 · LPIPS 0.145 · LSD 9.81

refwarm100k ep019 · PSNR 19.63 · LPIPS 0.152 · LSD 9.73

InstructAV2AV (first frame)

KLqhVxf4qGA — “running water”

removal (no first frame)

input

mainline ep018 · PSNR 25.33 · LPIPS 0.099 · LSD 11.27

refwarm100k ep018 · PSNR 26.44 · LPIPS 0.089 · LSD 11.51

refwarm100k ep019 · PSNR 26.69 · LPIPS 0.090 · LSD 12.04

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 23.10 · LPIPS 0.125 · LSD 10.01

refwarm100k ep018 · PSNR 25.54 · LPIPS 0.085 · LSD 9.31

refwarm100k ep019 · PSNR 25.27 · LPIPS 0.087 · LSD 10.45

InstructAV2AV (first frame)

Pl3ylTH0OeI — “mechanical hum”

removal (no first frame)

input

mainline ep018 · PSNR 20.47 · LPIPS 0.130 · LSD 11.71

refwarm100k ep018 · PSNR 16.93 · LPIPS 0.397 · LSD 13.73

refwarm100k ep019 · PSNR 18.61 · LPIPS 0.230 · LSD 9.66

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 18.60 · LPIPS 0.131 · LSD 15.37

refwarm100k ep018 · PSNR 20.27 · LPIPS 0.091 · LSD 13.81

refwarm100k ep019 · PSNR 20.04 · LPIPS 0.099 · LSD 14.41

InstructAV2AV (first frame)

qcOyrOSiprA — “stringed instrument”

removal (no first frame)

input

mainline ep018 · PSNR 20.97 · LPIPS 0.231 · LSD 11.07

refwarm100k ep018 · PSNR 22.75 · LPIPS 0.225 · LSD 11.05

refwarm100k ep019 · PSNR 22.93 · LPIPS 0.230 · LSD 11.13

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 16.10 · LPIPS 0.271 · LSD 11.00

refwarm100k ep018 · PSNR 17.55 · LPIPS 0.182 · LSD 10.60

refwarm100k ep019 · PSNR 17.33 · LPIPS 0.251 · LSD 10.71

InstructAV2AV (first frame)

ORJPtE5i2vY_000060 — “individuals”

removal (no first frame)

input

mainline ep018 · PSNR 34.24 · LPIPS 0.141 · LSD 10.33

refwarm100k ep018 · PSNR 33.86 · LPIPS 0.152 · LSD 10.72

refwarm100k ep019 · PSNR 26.68 · LPIPS 0.256 · LSD 11.00

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 21.88 · LPIPS 0.405 · LSD 10.17

refwarm100k ep018 · PSNR 25.73 · LPIPS 0.217 · LSD 10.04

refwarm100k ep019 · PSNR 25.32 · LPIPS 0.253 · LSD 9.86

InstructAV2AV (first frame)

52Muaxnp3mA — “animal”

removal (no first frame)

input

mainline ep018 · PSNR 15.85 · LPIPS 0.708 · LSD 15.50

refwarm100k ep018 · PSNR 16.95 · LPIPS 0.674 · LSD 15.29

refwarm100k ep019 · PSNR 17.33 · LPIPS 0.663 · LSD 16.12

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 14.60 · LPIPS 0.595 · LSD 10.48

refwarm100k ep018 · PSNR 13.50 · LPIPS 0.387 · LSD 10.34

refwarm100k ep019 · PSNR 13.53 · LPIPS 0.426 · LSD 10.64

InstructAV2AV (first frame)

EtWLI7oiEws_000045 — “tractor”

removal (no first frame)

input

mainline ep018 · PSNR 24.17 · LPIPS 0.224 · LSD 10.06

refwarm100k ep018 · PSNR 23.15 · LPIPS 0.264 · LSD 10.06

refwarm100k ep019 · PSNR 22.53 · LPIPS 0.255 · LSD 10.16

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.48 · LPIPS 0.290 · LSD 9.41

refwarm100k ep018 · PSNR 20.01 · LPIPS 0.195 · LSD 9.59

refwarm100k ep019 · PSNR 19.44 · LPIPS 0.214 · LSD 9.69

InstructAV2AV (first frame)

teofQ2tgfGk — “individual”

removal (no first frame)

input

mainline ep018 · PSNR 21.62 · LPIPS 0.232 · LSD 13.75

refwarm100k ep018 · PSNR 20.76 · LPIPS 0.247 · LSD 13.22

refwarm100k ep019 · PSNR 21.17 · LPIPS 0.255 · LSD 12.41

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 14.95 · LPIPS 0.339 · LSD 11.43

refwarm100k ep018 · PSNR 15.59 · LPIPS 0.302 · LSD 11.27

refwarm100k ep019 · PSNR 17.37 · LPIPS 0.282 · LSD 11.60

InstructAV2AV (first frame)

6bh9XSBN8aE — “string orchestra”

removal (no first frame)

input

mainline ep018 · PSNR 22.38 · LPIPS 0.338 · LSD 10.03

refwarm100k ep018 · PSNR 25.42 · LPIPS 0.212 · LSD 9.91

refwarm100k ep019 · PSNR 23.43 · LPIPS 0.293 · LSD 9.85

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 22.76 · LPIPS 0.184 · LSD 10.82

refwarm100k ep018 · PSNR 26.91 · LPIPS 0.068 · LSD 10.54

refwarm100k ep019 · PSNR 26.56 · LPIPS 0.070 · LSD 10.75

InstructAV2AV (first frame)

dCIGLbBwe4k_000250 — “trumpet”

removal (no first frame)

input

mainline ep018 · PSNR 20.43 · LPIPS 0.159 · LSD 15.11

refwarm100k ep018 · PSNR 19.19 · LPIPS 0.157 · LSD 14.67

refwarm100k ep019 · PSNR 20.84 · LPIPS 0.161 · LSD 14.73

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 13.78 · LPIPS 0.365 · LSD 15.13

refwarm100k ep018 · PSNR 16.02 · LPIPS 0.256 · LSD 15.34

refwarm100k ep019 · PSNR 15.74 · LPIPS 0.271 · LSD 15.28

InstructAV2AV (first frame)

gVdk8fEGL7c — “pianist”

removal (no first frame)

input

mainline ep018 · PSNR 22.91 · LPIPS 0.104 · LSD 11.25

refwarm100k ep018 · PSNR 15.80 · LPIPS 0.252 · LSD 11.17

refwarm100k ep019 · PSNR 22.03 · LPIPS 0.109 · LSD 11.74

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 19.82 · LPIPS 0.161 · LSD 9.88

refwarm100k ep018 · PSNR 26.24 · LPIPS 0.050 · LSD 10.15

refwarm100k ep019 · PSNR 26.13 · LPIPS 0.050 · LSD 9.89

InstructAV2AV (first frame)

CUrulzw7D-M_000050 — “crowd”

removal (no first frame)

input

mainline ep018 · PSNR 15.25 · LPIPS 0.608 · LSD 10.01

refwarm100k ep018 · PSNR 15.82 · LPIPS 0.497 · LSD 10.14

refwarm100k ep019 · PSNR 16.81 · LPIPS 0.364 · LSD 9.79

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.60 · LPIPS 0.335 · LSD 9.22

refwarm100k ep018 · PSNR 16.38 · LPIPS 0.296 · LSD 9.49

refwarm100k ep019 · PSNR 17.04 · LPIPS 0.295 · LSD 9.41

InstructAV2AV (first frame)

faFZzzkn5Qs_000030 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 22.59 · LPIPS 0.097 · LSD 26.65

refwarm100k ep018 · PSNR 22.60 · LPIPS 0.089 · LSD 24.32

refwarm100k ep019 · PSNR 22.18 · LPIPS 0.096 · LSD 25.22

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.97 · LPIPS 0.235 · LSD 11.75

refwarm100k ep018 · PSNR 19.57 · LPIPS 0.184 · LSD 11.63

refwarm100k ep019 · PSNR 19.44 · LPIPS 0.188 · LSD 12.40

InstructAV2AV (first frame)

-9HoMjD6Opk_000030 — “chainsaw”

removal (no first frame)

input

mainline ep018 · PSNR 22.00 · LPIPS 0.173 · LSD 10.03

refwarm100k ep018 · PSNR 21.60 · LPIPS 0.173 · LSD 12.32

refwarm100k ep019 · PSNR 21.48 · LPIPS 0.181 · LSD 13.40

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 18.92 · LPIPS 0.215 · LSD 11.73

refwarm100k ep018 · PSNR 23.96 · LPIPS 0.094 · LSD 13.11

refwarm100k ep019 · PSNR 23.29 · LPIPS 0.104 · LSD 12.03

InstructAV2AV (first frame)

QhbMdt7tBXc_000030 — “violin”

removal (no first frame)

input

mainline ep018 · PSNR 21.15 · LPIPS 0.200 · LSD 11.61

refwarm100k ep018 · PSNR 22.87 · LPIPS 0.180 · LSD 11.62

refwarm100k ep019 · PSNR 21.45 · LPIPS 0.191 · LSD 11.73

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.25 · LPIPS 0.289 · LSD 13.15

refwarm100k ep018 · PSNR 18.48 · LPIPS 0.210 · LSD 12.86

refwarm100k ep019 · PSNR 18.79 · LPIPS 0.210 · LSD 12.60

InstructAV2AV (first frame)

eaWRjpJ5f0w — “tool”

removal (no first frame)

input

mainline ep018 · PSNR 17.94 · LPIPS 0.303 · LSD 11.62

refwarm100k ep018 · PSNR 14.93 · LPIPS 0.433 · LSD 11.81

refwarm100k ep019 · PSNR 15.80 · LPIPS 0.446 · LSD 11.88

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 16.32 · LPIPS 0.263 · LSD 13.08

refwarm100k ep018 · PSNR 18.76 · LPIPS 0.163 · LSD 13.31

refwarm100k ep019 · PSNR 18.49 · LPIPS 0.169 · LSD 13.54

InstructAV2AV (first frame)

P6wEH1rgIm4_000178 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 18.92 · LPIPS 0.402 · LSD 11.43

refwarm100k ep018 · PSNR 18.80 · LPIPS 0.403 · LSD 11.58

refwarm100k ep019 · PSNR 18.90 · LPIPS 0.422 · LSD 18.84

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 10.96 · LPIPS 0.533 · LSD 10.08

refwarm100k ep018 · PSNR 9.70 · LPIPS 0.547 · LSD 10.20

refwarm100k ep019 · PSNR 9.83 · LPIPS 0.534 · LSD 9.80

InstructAV2AV (first frame)

4paU4HWkIag_000010 — “woman”

removal (no first frame)

input

mainline ep018 · PSNR 24.86 · LPIPS 0.140 · LSD 13.60

refwarm100k ep018 · PSNR 24.05 · LPIPS 0.143 · LSD 13.55

refwarm100k ep019 · PSNR 24.37 · LPIPS 0.138 · LSD 13.42

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 19.00 · LPIPS 0.182 · LSD 12.75

refwarm100k ep018 · PSNR 25.25 · LPIPS 0.052 · LSD 13.22

refwarm100k ep019 · PSNR 24.42 · LPIPS 0.059 · LSD 13.17

InstructAV2AV (first frame)

2rciuzaEEWo_000025 — “chimpanzee”

removal (no first frame)

input

mainline ep018 · PSNR 18.15 · LPIPS 0.313 · LSD 11.79

refwarm100k ep018 · PSNR 18.09 · LPIPS 0.294 · LSD 11.63

refwarm100k ep019 · PSNR 17.96 · LPIPS 0.319 · LSD 11.88

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.49 · LPIPS 0.293 · LSD 11.25

refwarm100k ep018 · PSNR 16.03 · LPIPS 0.274 · LSD 11.28

refwarm100k ep019 · PSNR 15.45 · LPIPS 0.300 · LSD 11.25

InstructAV2AV (first frame)

gmA7jC4XZAY_000030 — “string ensemble”

removal (no first frame)

input

mainline ep018 · PSNR 25.48 · LPIPS 0.078 · LSD 16.86

refwarm100k ep018 · PSNR 27.30 · LPIPS 0.067 · LSD 17.44

refwarm100k ep019 · PSNR 26.66 · LPIPS 0.071 · LSD 17.23

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 22.16 · LPIPS 0.094 · LSD 13.27

refwarm100k ep018 · PSNR 26.43 · LPIPS 0.048 · LSD 13.20

refwarm100k ep019 · PSNR 25.36 · LPIPS 0.056 · LSD 13.05

InstructAV2AV (first frame)

hHcLTDeq6Zk — “sewing machine”

removal (no first frame)

input

mainline ep018 · PSNR 10.95 · LPIPS 0.751 · LSD 21.53

refwarm100k ep018 · PSNR 11.76 · LPIPS 0.782 · LSD 21.49

refwarm100k ep019 · PSNR 10.96 · LPIPS 0.776 · LSD 16.68

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 9.40 · LPIPS 0.609 · LSD 10.55

refwarm100k ep018 · PSNR 12.12 · LPIPS 0.452 · LSD 10.72

refwarm100k ep019 · PSNR 11.70 · LPIPS 0.461 · LSD 10.56

InstructAV2AV (first frame)

z5MSCqZ3pc0_000040 — “sewing machine”

removal (no first frame)

input

mainline ep018 · PSNR 19.99 · LPIPS 0.252 · LSD 12.54

refwarm100k ep018 · PSNR 23.46 · LPIPS 0.187 · LSD 12.39

refwarm100k ep019 · PSNR 19.78 · LPIPS 0.263 · LSD 12.11

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.04 · LPIPS 0.279 · LSD 11.65

refwarm100k ep018 · PSNR 17.66 · LPIPS 0.248 · LSD 11.14

refwarm100k ep019 · PSNR 17.26 · LPIPS 0.287 · LSD 12.64

InstructAV2AV (first frame)

77Rlbds5MEU_000261 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 21.91 · LPIPS 0.110 · LSD 13.99

refwarm100k ep018 · PSNR 21.64 · LPIPS 0.219 · LSD 17.73

refwarm100k ep019 · PSNR 22.97 · LPIPS 0.154 · LSD 18.23

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 21.22 · LPIPS 0.149 · LSD 24.40

refwarm100k ep018 · PSNR 22.88 · LPIPS 0.112 · LSD 24.62

refwarm100k ep019 · PSNR 22.43 · LPIPS 0.134 · LSD 24.30

InstructAV2AV (first frame)

J7j7Rq7jb0A_000570 — “watercraft”

removal (no first frame)

input

mainline ep018 · PSNR 23.10 · LPIPS 0.275 · LSD 22.07

refwarm100k ep018 · PSNR 23.07 · LPIPS 0.281 · LSD 20.46

refwarm100k ep019 · PSNR 23.08 · LPIPS 0.305 · LSD 21.41

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 19.67 · LPIPS 0.342 · LSD 10.78

refwarm100k ep018 · PSNR 20.26 · LPIPS 0.271 · LSD 10.14

refwarm100k ep019 · PSNR 20.00 · LPIPS 0.257 · LSD 10.66

InstructAV2AV (first frame)

fNsb_BdIHhI — “waves”

removal (no first frame)

input

mainline ep018 · PSNR 20.53 · LPIPS 0.366 · LSD 11.05

refwarm100k ep018 · PSNR 20.08 · LPIPS 0.370 · LSD 10.58

refwarm100k ep019 · PSNR 19.14 · LPIPS 0.387 · LSD 16.33

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.43 · LPIPS 0.496 · LSD 8.94

refwarm100k ep018 · PSNR 17.99 · LPIPS 0.452 · LSD 8.92

refwarm100k ep019 · PSNR 17.99 · LPIPS 0.440 · LSD 9.83

InstructAV2AV (first frame)

CYQDPwKvchg_000030 — “large transport vehicle”

removal (no first frame)

input

mainline ep018 · PSNR 13.79 · LPIPS 0.307 · LSD 22.03

refwarm100k ep018 · PSNR 14.32 · LPIPS 0.313 · LSD 22.34

refwarm100k ep019 · PSNR 13.82 · LPIPS 0.312 · LSD 22.70

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.94 · LPIPS 0.252 · LSD 12.78

refwarm100k ep018 · PSNR 18.53 · LPIPS 0.193 · LSD 12.72

refwarm100k ep019 · PSNR 18.11 · LPIPS 0.210 · LSD 12.31

InstructAV2AV (first frame)

ZRbAhRimdGU — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 18.63 · LPIPS 0.452 · LSD 16.11

refwarm100k ep018 · PSNR 18.56 · LPIPS 0.453 · LSD 14.80

refwarm100k ep019 · PSNR 18.95 · LPIPS 0.469 · LSD 17.13

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 14.89 · LPIPS 0.432 · LSD 13.28

refwarm100k ep018 · PSNR 15.84 · LPIPS 0.365 · LSD 13.24

refwarm100k ep019 · PSNR 15.76 · LPIPS 0.360 · LSD 12.81

InstructAV2AV (first frame)

GFPYtMEZIsU — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 19.78 · LPIPS 0.382 · LSD 10.61

refwarm100k ep018 · PSNR 19.72 · LPIPS 0.404 · LSD 10.77

refwarm100k ep019 · PSNR 19.10 · LPIPS 0.418 · LSD 10.50

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.83 · LPIPS 0.465 · LSD 10.99

refwarm100k ep018 · PSNR 17.10 · LPIPS 0.399 · LSD 11.31

refwarm100k ep019 · PSNR 17.00 · LPIPS 0.410 · LSD 10.88

InstructAV2AV (first frame)

gD0wNBWQLjo_000108 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 21.07 · LPIPS 0.234 · LSD 8.86

refwarm100k ep018 · PSNR 14.99 · LPIPS 0.259 · LSD 8.85

refwarm100k ep019 · PSNR 26.51 · LPIPS 0.146 · LSD 8.91

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.00 · LPIPS 0.193 · LSD 14.95

refwarm100k ep018 · PSNR 19.38 · LPIPS 0.139 · LSD 14.53

refwarm100k ep019 · PSNR 18.68 · LPIPS 0.152 · LSD 14.37

InstructAV2AV (first frame)

UQxlzZS7x3k_000050 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 20.69 · LPIPS 0.187 · LSD 8.85

refwarm100k ep018 · PSNR 20.96 · LPIPS 0.198 · LSD 8.91

refwarm100k ep019 · PSNR 20.80 · LPIPS 0.183 · LSD 8.92

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 25.07 · LPIPS 0.149 · LSD 19.69

refwarm100k ep018 · PSNR 30.15 · LPIPS 0.064 · LSD 18.15

refwarm100k ep019 · PSNR 29.93 · LPIPS 0.071 · LSD 14.60

InstructAV2AV (first frame)

GD_gkamXLLU — “individual”

removal (no first frame)

input

mainline ep018 · PSNR 15.84 · LPIPS 0.399 · LSD 13.16

refwarm100k ep018 · PSNR 15.44 · LPIPS 0.396 · LSD 13.12

refwarm100k ep019 · PSNR 15.18 · LPIPS 0.410 · LSD 13.32

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 13.09 · LPIPS 0.485 · LSD 13.62

refwarm100k ep018 · PSNR 17.26 · LPIPS 0.222 · LSD 13.48

refwarm100k ep019 · PSNR 18.26 · LPIPS 0.194 · LSD 13.48

InstructAV2AV (first frame)

ZhBpATtRZzE_000187 — “trumpet”

removal (no first frame)

input

mainline ep018 · PSNR 24.26 · LPIPS 0.064 · LSD 32.63

refwarm100k ep018 · PSNR 25.40 · LPIPS 0.052 · LSD 31.38

refwarm100k ep019 · PSNR 23.42 · LPIPS 0.073 · LSD 32.59

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 16.07 · LPIPS 0.234 · LSD 17.24

refwarm100k ep018 · PSNR 22.02 · LPIPS 0.056 · LSD 15.74

refwarm100k ep019 · PSNR 22.07 · LPIPS 0.058 · LSD 17.64

InstructAV2AV (first frame)

h8D7I5aAZEc_000070 — “brass band”

removal (no first frame)

input

mainline ep018 · PSNR 21.47 · LPIPS 0.337 · LSD 12.27

refwarm100k ep018 · PSNR 20.58 · LPIPS 0.346 · LSD 12.36

refwarm100k ep019 · PSNR 20.46 · LPIPS 0.387 · LSD 12.68

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.53 · LPIPS 0.366 · LSD 13.83

refwarm100k ep018 · PSNR 16.33 · LPIPS 0.314 · LSD 13.78

refwarm100k ep019 · PSNR 16.41 · LPIPS 0.329 · LSD 13.31

InstructAV2AV (first frame)

bBZvoYapffk_000007 — “golf club”

removal (no first frame)

input

mainline ep018 · PSNR 21.48 · LPIPS 0.199 · LSD 9.51

refwarm100k ep018 · PSNR 21.68 · LPIPS 0.197 · LSD 9.72

refwarm100k ep019 · PSNR 21.55 · LPIPS 0.179 · LSD 9.64

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 25.97 · LPIPS 0.072 · LSD 14.71

refwarm100k ep018 · PSNR 26.36 · LPIPS 0.058 · LSD 14.11

refwarm100k ep019 · PSNR 26.42 · LPIPS 0.061 · LSD 9.91

InstructAV2AV (first frame)

AmXIsa0ED9g — “car engine”

removal (no first frame)

input

mainline ep018 · PSNR 21.93 · LPIPS 0.308 · LSD 12.69

refwarm100k ep018 · PSNR 22.45 · LPIPS 0.305 · LSD 12.22

refwarm100k ep019 · PSNR 22.32 · LPIPS 0.307 · LSD 12.97

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.54 · LPIPS 0.423 · LSD 10.62

refwarm100k ep018 · PSNR 16.95 · LPIPS 0.422 · LSD 10.03

refwarm100k ep019 · PSNR 16.94 · LPIPS 0.388 · LSD 10.72

InstructAV2AV (first frame)

v4lX1z_noGI — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 23.75 · LPIPS 0.153 · LSD 10.63

refwarm100k ep018 · PSNR 23.16 · LPIPS 0.162 · LSD 10.79

refwarm100k ep019 · PSNR 22.45 · LPIPS 0.195 · LSD 10.06

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.09 · LPIPS 0.268 · LSD 10.62

refwarm100k ep018 · PSNR 19.69 · LPIPS 0.197 · LSD 10.56

refwarm100k ep019 · PSNR 19.88 · LPIPS 0.187 · LSD 10.59

InstructAV2AV (first frame)

ou0qx4s0Tbw — “electronic device”

removal (no first frame)

input

mainline ep018 · PSNR 16.85 · LPIPS 0.441 · LSD 11.60

refwarm100k ep018 · PSNR 15.70 · LPIPS 0.535 · LSD 10.93

refwarm100k ep019 · PSNR 19.88 · LPIPS 0.294 · LSD 9.62

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 9.83 · LPIPS 0.630 · LSD 15.84

refwarm100k ep018 · PSNR 27.88 · LPIPS 0.057 · LSD 15.92

refwarm100k ep019 · PSNR 27.03 · LPIPS 0.059 · LSD 15.79

InstructAV2AV (first frame)

YBdCF5qm4s0 — “train”

removal (no first frame)

input

mainline ep018 · PSNR 19.45 · LPIPS 0.151 · LSD 13.59

refwarm100k ep018 · PSNR 19.52 · LPIPS 0.149 · LSD 16.23

refwarm100k ep019 · PSNR 19.53 · LPIPS 0.143 · LSD 13.84

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.18 · LPIPS 0.161 · LSD 10.73

refwarm100k ep018 · PSNR 16.53 · LPIPS 0.165 · LSD 10.44

refwarm100k ep019 · PSNR 16.87 · LPIPS 0.158 · LSD 12.42

InstructAV2AV (first frame)

BU2cn5cISlU — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 21.21 · LPIPS 0.137 · LSD 11.45

refwarm100k ep018 · PSNR 21.07 · LPIPS 0.146 · LSD 11.47

refwarm100k ep019 · PSNR 21.13 · LPIPS 0.145 · LSD 11.82

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 18.56 · LPIPS 0.203 · LSD 10.44

refwarm100k ep018 · PSNR 19.82 · LPIPS 0.157 · LSD 10.24

refwarm100k ep019 · PSNR 19.46 · LPIPS 0.171 · LSD 10.22

InstructAV2AV (first frame)

uOrvpS1ueyk — “rooster”

removal (no first frame)

input

mainline ep018 · PSNR 19.10 · LPIPS 0.508 · LSD 18.95

refwarm100k ep018 · PSNR 19.34 · LPIPS 0.472 · LSD 18.35

refwarm100k ep019 · PSNR 19.58 · LPIPS 0.463 · LSD 20.85

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.63 · LPIPS 0.306 · LSD 12.84

refwarm100k ep018 · PSNR 17.69 · LPIPS 0.232 · LSD 12.76

refwarm100k ep019 · PSNR 18.56 · LPIPS 0.208 · LSD 13.56

InstructAV2AV (first frame)

ZwvwMfyjzdE — “fire engine”

removal (no first frame)

input

mainline ep018 · PSNR 16.56 · LPIPS 0.403 · LSD 13.33

refwarm100k ep018 · PSNR 16.75 · LPIPS 0.377 · LSD 14.96

refwarm100k ep019 · PSNR 16.48 · LPIPS 0.349 · LSD 13.70

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 11.49 · LPIPS 0.497 · LSD 10.10

refwarm100k ep018 · PSNR 11.80 · LPIPS 0.453 · LSD 10.11

refwarm100k ep019 · PSNR 11.87 · LPIPS 0.448 · LSD 9.94

InstructAV2AV (first frame)

Oloub-M4bxs — “crowd”

removal (no first frame)

input

mainline ep018 · PSNR 13.16 · LPIPS 0.666 · LSD 17.15

refwarm100k ep018 · PSNR 12.90 · LPIPS 0.677 · LSD 25.12

refwarm100k ep019 · PSNR 13.19 · LPIPS 0.704 · LSD 10.57

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 14.50 · LPIPS 0.253 · LSD 9.77

refwarm100k ep018 · PSNR 14.68 · LPIPS 0.231 · LSD 9.66

refwarm100k ep019 · PSNR 14.71 · LPIPS 0.228 · LSD 9.75

InstructAV2AV (first frame)

NvoqO63JNZk — “vehicle engine”

removal (no first frame)

input

mainline ep018 · PSNR 14.00 · LPIPS 0.634 · LSD 11.07

refwarm100k ep018 · PSNR 12.75 · LPIPS 0.704 · LSD 11.18

refwarm100k ep019 · PSNR 13.75 · LPIPS 0.705 · LSD 10.63

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 16.56 · LPIPS 0.338 · LSD 9.38

refwarm100k ep018 · PSNR 16.78 · LPIPS 0.289 · LSD 9.28

refwarm100k ep019 · PSNR 16.30 · LPIPS 0.308 · LSD 9.32

InstructAV2AV (first frame)

55M-kgQjip0_000530 — “trumpet”

removal (no first frame)

input

mainline ep018 · PSNR 17.45 · LPIPS 0.185 · LSD 9.74

refwarm100k ep018 · PSNR 15.03 · LPIPS 0.218 · LSD 10.14

refwarm100k ep019 · PSNR 14.58 · LPIPS 0.220 · LSD 10.12

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 17.05 · LPIPS 0.153 · LSD 10.14

refwarm100k ep018 · PSNR 14.70 · LPIPS 0.203 · LSD 10.12

refwarm100k ep019 · PSNR 18.74 · LPIPS 0.116 · LSD 10.12

InstructAV2AV (first frame)

ivZnUkicVkU_000093 — “dog”

removal (no first frame)

input

mainline ep018 · PSNR 20.52 · LPIPS 0.269 · LSD 13.58

refwarm100k ep018 · PSNR 22.56 · LPIPS 0.184 · LSD 11.05

refwarm100k ep019 · PSNR 19.73 · LPIPS 0.344 · LSD 11.07

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 12.06 · LPIPS 0.333 · LSD 11.83

refwarm100k ep018 · PSNR 13.46 · LPIPS 0.309 · LSD 11.47

refwarm100k ep019 · PSNR 13.51 · LPIPS 0.316 · LSD 11.80

InstructAV2AV (first frame)

Po6ZZ0sGfUw_000011 — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 24.36 · LPIPS 0.133 · LSD 10.13

refwarm100k ep018 · PSNR 23.44 · LPIPS 0.151 · LSD 10.07

refwarm100k ep019 · PSNR 16.87 · LPIPS 0.544 · LSD 10.66

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 19.07 · LPIPS 0.201 · LSD 10.93

refwarm100k ep018 · PSNR 20.68 · LPIPS 0.198 · LSD 10.56

refwarm100k ep019 · PSNR 20.69 · LPIPS 0.201 · LSD 10.83

InstructAV2AV (first frame)

At2fuS2bpbA_000049 — “violin”

removal (no first frame)

input

mainline ep018 · PSNR 18.13 · LPIPS 0.318 · LSD 13.41

refwarm100k ep018 · PSNR 17.45 · LPIPS 0.358 · LSD 17.24

refwarm100k ep019 · PSNR 16.83 · LPIPS 0.299 · LSD 15.44

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 13.83 · LPIPS 0.402 · LSD 12.65

refwarm100k ep018 · PSNR 14.98 · LPIPS 0.298 · LSD 13.68

refwarm100k ep019 · PSNR 15.02 · LPIPS 0.303 · LSD 13.51

InstructAV2AV (first frame)

bfME7zcB2BY — “birds”

removal (no first frame)

input

mainline ep018 · PSNR 26.34 · LPIPS 0.168 · LSD 20.54

refwarm100k ep018 · PSNR 26.13 · LPIPS 0.172 · LSD 22.13

refwarm100k ep019 · PSNR 26.08 · LPIPS 0.171 · LSD 11.04

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 13.40 · LPIPS 0.428 · LSD 9.98

refwarm100k ep018 · PSNR 17.90 · LPIPS 0.320 · LSD 9.97

refwarm100k ep019 · PSNR 18.18 · LPIPS 0.305 · LSD 9.70

InstructAV2AV (first frame)

_4C-JDyGhSw_000030 — “guitar”

removal (no first frame)

input

mainline ep018 · PSNR 20.89 · LPIPS 0.203 · LSD 38.94

refwarm100k ep018 · PSNR 18.03 · LPIPS 0.295 · LSD 40.46

refwarm100k ep019 · PSNR 19.67 · LPIPS 0.231 · LSD 36.26

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 14.85 · LPIPS 0.297 · LSD 20.48

refwarm100k ep018 · PSNR 17.67 · LPIPS 0.193 · LSD 19.94

refwarm100k ep019 · PSNR 17.70 · LPIPS 0.197 · LSD 23.98

InstructAV2AV (first frame)

KihiqmGcJG0_000030 — “synthesizer”

removal (no first frame)

input

mainline ep018 · PSNR 20.51 · LPIPS 0.254 · LSD 18.01

refwarm100k ep018 · PSNR 19.54 · LPIPS 0.279 · LSD 18.19

refwarm100k ep019 · PSNR 20.99 · LPIPS 0.153 · LSD 17.74

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 15.22 · LPIPS 0.293 · LSD 14.47

refwarm100k ep018 · PSNR 15.31 · LPIPS 0.300 · LSD 14.44

refwarm100k ep019 · PSNR 16.35 · LPIPS 0.260 · LSD 14.49

InstructAV2AV (first frame)

AftwRi6r71c — “female vocalist”

removal (no first frame)

input

mainline ep018 · PSNR 22.17 · LPIPS 0.190 · LSD 10.40

refwarm100k ep018 · PSNR 21.92 · LPIPS 0.206 · LSD 10.52

refwarm100k ep019 · PSNR 21.89 · LPIPS 0.208 · LSD 10.22

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 20.11 · LPIPS 0.239 · LSD 10.60

refwarm100k ep018 · PSNR 25.86 · LPIPS 0.111 · LSD 10.30

refwarm100k ep019 · PSNR 26.20 · LPIPS 0.116 · LSD 10.34

InstructAV2AV (first frame)

54AM8JU1e9o_000030 — “electric guitar”

removal (no first frame)

input

mainline ep018 · PSNR 22.08 · LPIPS 0.235 · LSD 12.31

refwarm100k ep018 · PSNR 21.00 · LPIPS 0.254 · LSD 12.08

refwarm100k ep019 · PSNR 21.81 · LPIPS 0.235 · LSD 12.24

InstructAV2AV

addition (first frame)

input

mainline ep018 · PSNR 14.27 · LPIPS 0.376 · LSD 11.43

refwarm100k ep018 · PSNR 18.38 · LPIPS 0.166 · LSD 11.42

refwarm100k ep019 · PSNR 18.60 · LPIPS 0.187 · LSD 11.38

InstructAV2AV (first frame)