Log in

View Full Version : madVR - high quality video renderer (GPU assisted)


Pages : 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 [736] 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329

ryrynz
9th March 2016, 02:17
In all other doubling / downscaling scenarios, 2d works fine (as does jinc), only adds around 2-3ms to render times. When supersampling render times shoot up into the 80s ms with 2d, from around 35ms with 1d. Jinc downscaling also shows the same behaviour when supersampling, its in the osd twice and render times shoot up into the 80s.

I think you may underestimate how demanding Jinc and AR are, only madshi can comment on the performance and OSD though really. No problems have been reported about the performance of Jinc or AR, so I expect that everything is fine here.

iSeries
9th March 2016, 02:22
I think you may underestimate how demanding Jinc and AR are, only madshi can comment on the performance and OSD though really. No problems have been reported about the performance of Jinc or AR, so I expect that everything is fine here.

If it was so demanding it would be the same downscaling 4k to 1080p, but with the 4k samples i have i can use 2d ssim downscaling absolutely fine (GTX950), well below maximum render times (and its only in the osd once). This issue only happens with supersampling.

CarlosCaco
9th March 2016, 02:23
What does the "present a frame for every vsync" option have to do with this?






i was just saying that this option increase a little my render times, without this option checked i having some glitches...

ryrynz
9th March 2016, 02:30
If it was so demanding it would be the same downscaling 4k to 1080p, but with the 4k samples i have i can use 2d ssim downscaling absolutely fine (GTX950), well below maximum render times (and its only in the osd once). This issue only happens with supersampling.

No it wouldn't, you're not factoring in image doubling.

madshi, this will probably be addressed regarding the OSD, but Jinc AR shows up in the OSD when supersampling and downscaling using Jinc with no AR. It shows Image doubler = Jinc AR < Jinc.

iSeries
9th March 2016, 02:36
No it wouldn't, you're not factoring in image doubling.

madshi, this will probably be addressed regarding the OSD, but Jinc AR shows up in the OSD when supersampling and downscaling using Jinc with no AR. It shows Image doubler = Jinc AR < Jinc.

When supersampling, doubling 1080p and downscaling back to 1080p with 1d, using super-xbr to double, and using superres 3 and crispen edges to enhance, i get around 35ms render times. Downscaling with 2d it shoots up into the 80s. That can't be normal. Like i said, all other doubling scenarios 2d only adds around 2-3ms to what 1d costs, and same goes for downscaling 4k native to 1080p, no trouble at all using 2d, again only adds 2-3ms to what 1d costs.

ryrynz
9th March 2016, 03:13
When supersampling, doubling 1080p and downscaling back to 1080p with 1d, using super-xbr to double, and using superres 3 and crispen edges to enhance, i get around 35ms render times. Downscaling with 2d it shoots up into the 80s. That can't be normal. Like i said, all other doubling scenarios 2d only adds around 2-3ms to what 1d costs, and same goes for downscaling 4k native to 1080p, no trouble at all using 2d, again only adds 2-3ms to what 1d costs.

Only 2ms difference between 1D and 2D? I think something's broken, you should maybe look into resetting your settings as my GTX 960 has about 8ms render times when downscaling 4K content and 16-18ms when using 2D (with AR and LL)

tp4tissue
9th March 2016, 07:11
Hey guys.. so , with Madvr.. If I want to use it, I have to stick with the 32-bit.

What do I use to decode 10bit hevc 4K to then connect with Madvr, because my downstairs computer is not fast enough using the 32bit Lav filter to CPU decode that.

I'm a Madvr diehard.. I just thought this would be the place to post this, since you guys are also sticking with 32bit for madvr..

huhn
9th March 2016, 07:19
madVR has a 64 bit version for a very long time so no a lot of us are using 64 bit.

tp4tissue
9th March 2016, 08:09
madVR has a 64 bit version for a very long time so no a lot of us are using 64 bit.


what.. what............................ wait, is it the same download in the OP ?

ryrynz
9th March 2016, 08:37
what.. what............................ wait, is it the same download in the OP ?

The zip contains both the 32 bit version and the 64 bit version. Make sure to run install.bat after extraction.

ileile
9th March 2016, 18:08
Sometimes stereo 3d playback is not smooth, and the "dropped frames" or "presentation glitches" keep increasing. My CPU is not good enough (i3 2100), but I think it is not the reason. Because I tried some times with same movie clip, but the problem is not easy to reproduce.

another problem is that: if I connect two displays and play something using mpc-hc, when I drag the mpc-hc from one display to another, the system often crash. Not BSOD, but something like black screen, system can not respond(ctrl+alt+del usually can't work too, but I can move mouse pointer for a period). this problem may occur with only one display, when switching between window mode and exclusive mode. My GPU is AMD R7 200.

system is windows 10.

chros
9th March 2016, 20:00
what.. what............................ wait, is it the same download in the OP ?
:) yep, yep, I haven't realized it for a long time either, just kept extracting the zip file :)

James Freeman
9th March 2016, 20:19
madshi can you please add a ST.2084 EOTF transfer function under calibration tab?

aufkrawall
10th March 2016, 00:22
Sometimes my GTX 980 gets stuck in the wrong powerstate when opening videos, it doesn't run with full clock then and drops lots of frames. I suspect this is a randomly triggered bug by NNEDI3 in the OpenCL driver, a bit annoying.
Has someone else observed anything similar? Happens with the current driver (364.51) and some previous ones as well. Win 10 x64.

fedpul
10th March 2016, 00:25
Only 2ms difference between 1D and 2D? I think something's broken, you should maybe look into resetting your settings as my GTX 960 has about 8ms render times when downscaling 4K content and 16-18ms when using 2D (with AR and LL)

Hi, I can confirm that supersampling 720, the difference between SSIM 1D and SSIM 2D is between 2-3 ms. I have a GTX 970 but I can also see that when supersampling 1080 the change in rendering times are kind of abnormal, it is a jump of 30-40 ms between 1D and 2D the same haapens with Jinc. I'm not sure if this is normal or not.

I'm also trying to detect what is causing black screens in 90.15. I'm almost sure that is NNEDI in Chroma Upscaling but I need to do some more testing.

fedpul
10th March 2016, 00:27
Sometimes my GTX 980 gets stuck in the wrong powerstate when opening videos, it doesn't run with full clock then and drops lots of frames. I suspect this is a randomly triggered bug by NNEDI3 in the OpenCL driver, a bit annoying.
Has someone else observed anything similar? Happens with the current driver (364.51) and some previous ones as well. Win 10 x64.

It happens to me when I update madvr and try to open a movie without restarting the PC.

bcec
10th March 2016, 00:34
is superxbr + sr4 preferable over nnedi3@64 + sr3 when doing 1080p->2160p upscaling?

I seem to have slight preference to the latter, but I am not sure if it is placebo, or if there are indeed noticeable difference making the latter a better option?

ryrynz
10th March 2016, 01:24
is superxbr + sr4 preferable over nnedi3@64 + sr3 when doing 1080p->2160p upscaling?

I seem to have slight preference to the latter, but I am not sure if it is placebo, or if there are indeed noticeable difference making the latter a better option?

Does it really matter what the other users think? It's your opinion that matters more. Screenshot and compare to see if there's a noticeable difference. Everyone has their own preferences.

Ver Greeneyes
10th March 2016, 01:39
is superxbr + sr4 preferable over nnedi3@64 + sr3 when doing 1080p->2160p upscaling?
I actually tend to stick to sr2 - I find anything higher to be too harsh. But that's pretty much personal preference.

QBhd
10th March 2016, 01:46
madshi, has the logic behind when doubling occurs changed? I can no longer use NNEDI3 to double the luma y when I play back 1280x720 ==> 1024x768. I was really happy with the results when I was able to do this:

http://i.imgur.com/m3hTBoW.png

With 90.15 this is what is happening:

http://i.imgur.com/lxizcHQ.png

This time I have included my settings.bin which you asked for the last time.

settings.bin (https://mega.nz/#!kFAVXRpR!3PUCCyiCW4-jqVECzEECYU2v-WgNeR7ABJgihEbEXJ0)

fedpul
10th March 2016, 02:30
madshi, has the logic behind when doubling occurs changed? I can no longer use NNEDI3 to double the luma y when I play back 1280x720 ==> 1024x768. I was really happy with the results when I was able to do this:

http://i.imgur.com/m3hTBoW.png

With 90.15 this is what is happening:

http://i.imgur.com/lxizcHQ.png

This time I have included my settings.bin which you asked for the last time.

settings.bin (https://mega.nz/#!kFAVXRpR!3PUCCyiCW4-jqVECzEECYU2v-WgNeR7ABJgihEbEXJ0)

That is beacuse in the first picture you are doubling just luma y (vertical resolución if I am not wrong) and in the second you must be using supersampling so both x and y are doubled. But indeed those settings seems overkill. You should be fine with using nnedi 32 for chroma and 64 for doubling luma. Hope this helps.

QBhd
10th March 2016, 02:54
It's the exact same settings... just different versions of madVR. I'm not worried about overkill (noise and heat are non-existant and I don't pay for electricity)... The problem is with the new logic, I can't use anything that doubles the image, it's far too demanding. I don't need to double the x. 1280>2560<1024 is not something I want or need. Doubling the y is rather nice 720>1440<768 is a thing of beauty (to me at least)

QB

fedpul
10th March 2016, 03:18
Your point is good, maybe in the future could be an option for selecting to double just y, x or both.

Enviado desde mi MI 4LTE mediante Tapatalk

bcec
10th March 2016, 04:27
Does it really matter what the other users think? It's your opinion that matters more. Screenshot and compare to see if there's a noticeable difference. Everyone has their own preferences.

Great suggestion! what's the right way to take a screen shot in madvr. I see that printscreen gives me a black screen

Asmodian
10th March 2016, 04:31
You have to turn off "Windowed overlay" to take screenshots.

bcec
10th March 2016, 04:52
You have to turn off "Windowed overlay" to take screenshots.

Thank you!

bcec
10th March 2016, 04:59
Does it really matter what the other users think? It's your opinion that matters more. Screenshot and compare to see if there's a noticeable difference. Everyone has their own preferences.

Thanks for the recommendation!

I actually tend to stick to sr2 - I find anything higher to be too harsh. But that's pretty much personal preference.

After comparing screenshots, you are absolutely right, 4 is quite harsh. I like 2 as well.

[edited]: Also, super-xbr seem to resolve more aliasing than nnedi3@64. Look at the left round frame of his eyeglasses, super-xbr is doing a much better job:

Nnedi3@64:
http://s11.postimg.org/a4xbf8cur/vic_nedi_2.png

Super-xbr:
http://s11.postimg.org/aiynetwyb/vic_xbr_2.png

leandronb
10th March 2016, 05:02
It could be a problem with the video decoder, but it's hard to say for sure. Which video decoder are you using? LAV? Which decoding mode (software, DXVA native, DXVA copyback, CUVID, ...)?

The software i am using to play video:
MPC-HC + LAV filters + AC3 filter + madvr.
And the decoding mode is CUVID, using a gtx 660ti.

ryrynz
10th March 2016, 05:11
Also, superxbr seem to resolve less aliasing than nnedi3@64. Look at the left round frame of his eyeglasses:


I think what you meant to say was that suber-xbr resolves more aliasing than NNEDI3 64 neurons. Super-xbr is looking good there, although the sizes of the images aren't the same.
The top of the ear is another noteworthy mention for suber-xbr here also, NNEDI3 dominates for lower resolution animated content without question though.

bcec
10th March 2016, 05:22
I think what you meant to say was that suber-xbr resolves more aliasing than NNEDI3 64 neurons. Super-xbr is looking good there, although the sizes of the images aren't the same.

yup, that's what I meant (edited the post) :thanks:

btw, the size of the original images are the same. they are just not cropped equally (i did a quick crop of the face before posting)

Warner306
10th March 2016, 06:49
is superxbr + sr4 preferable over nnedi3@64 + sr3 when doing 1080p->2160p upscaling?

I seem to have slight preference to the latter, but I am not sure if it is placebo, or if there are indeed noticeable difference making the latter a better option?

I prefer SuperRes (3) at a 2x scaling factor. I don't like an oversharpened image, but find I can improve the image further by adding crispen edges (1.0), as well. This will make higher SuperRes values look better.

super-xbr and NNEDI3 look similar with SuperRes, but super-xbr will enhance any existing ringing, while NNEDI3 will not. If you can't see ringing, or if doesn't bother you, use super-xbr.

RyuzakiL
10th March 2016, 08:10
Has anyone tried the latest Crimson driver 16.3 on Win10x64, MadVR?

and does MadVR work properly on Win10x64? or is it still Win8.1 that MadVR works perfectly.

ryrynz
10th March 2016, 08:51
Has anyone tried the latest Crimson driver 16.3 on Win10x64, MadVR?

and does MadVR work properly on Win10x64? or is it still Win8.1 that MadVR works perfectly.

Why not test it yourself?

Windows 10 works perfectly for me using a GTX 960 with latest everything.

rxracer
10th March 2016, 09:27
Sorry if this has been asked before. 1840 pages was daunting, and I couldn't find an answer searching.

I have 2 displays. If I set the viewing display to secondary I get dropped frames and glitches. If I'm doing nothing on the PC it drops a frame and produces a glitch every 7 seconds on the dot. I timed it.. bizarre. If I start using the pc on the primary monitor the video drops frames and glitches rapidly.

Is any of this a known issue? I've changed and disabled every single setting in MadVR, to no avail. I use Lav Filters and Zoom Player. Thinking about it now I'll try with MPC-HC. I hope it isn''t the player cos I really don't like any others.

huhn
10th March 2016, 10:08
Why not test it yourself?

Windows 10 works perfectly for me using a GTX 960 with latest everything.

there are problem with deinterlancing with windows 10.

of cause this is not an madVR error.

retrue
10th March 2016, 11:49
I am using last development versions of MPC-HC and LAVFilters.
I updated yesterday from madVR 0.9010 to 0.9015. When I try to play videos I get sound but not image, not in window mode, neither in fullscreen (curiously, with the last version of PotPlayer+MadVR I have image in fullscreen mode but not in window mode). This happens with vids at 1080p and codec H264. It doesn't happen with vids at 480p and codec H264.
The problem appears with version 0.9011. I have returned to madVR 0.9010 and the problem has disappeared.

huhn
10th March 2016, 13:36
are you using image doubling with "always"?

you may just run out of vram.

retrue
10th March 2016, 14:04
are you using image doubling with "always"?

you may just run out of vram.

Image doubling: double luma (64 neurons) and double chroma (32 neurons), always if upscaling is needed.
But with version madvr 0.9010 I don't have problems.

adhara
10th March 2016, 14:07
Hi,

When doing 1080p to 2160p upscale with MadVr, I get some video judders (every 15/20 sec) on my UHD TV (2160p/60Hz & HDCP 2.2).
The video input signal on my TV is well seen as 2160p.
I tested several video rates: 23.976Hz, 24 Hz, ... Same issue.

No dropped/delayed/... frame on MadVr. My GC is a Nvidia 970 GTX.

How come ?

When playing back a 2160p video directly on the TV (with an usb key for example), no problem.

fedpul
10th March 2016, 14:34
Hi, I can confirm that supersampling 720, the difference between SSIM 1D and SSIM 2D is between 2-3 ms. I have a GTX 970 but I can also see that when supersampling 1080 the change in rendering times are kind of abnormal, it is a jump of 30-40 ms between 1D and 2D the same haapens with Jinc. I'm not sure if this is normal or not.

I'm also trying to detect what is causing black screens in 90.15. I'm almost sure that is NNEDI in Chroma Upscaling but I need to do some more testing.

Quoting myself, I have done more tests.

Hi madshi, I kept getting black screens with 90.15, with Chroma NNEDI enabled, Doublind disabled, Chroma Jinc and Doubling NNEDI, and obviously with both using NNEDI. I need to check without using NNEDI at all. I always use NNEDI 32 Neurons for the tests. Black screens are really random, first one occurs after more than 50 minutes of playback, so testing requires lots of time.

nevcairiel
10th March 2016, 15:13
non-framepacked 3D output is broken in 0.90.15.

SBS only has the right eye showing, TB only the bottom, line interleaved is entirely black, and column interleaved behaves just like SBS (which seems odd in itself). On top of that, if i tick the swap eye checkbox, it becomes entirely black all the time.
When I pause playback the image shows up properly, but goes away again when I play again, at least for SBS and TB, interleaved seems to always stay dead.

On the same note, it would be nice if "auto" 3D mode would not switch the resolution if the OS doesn't even offer a stereo 3d mode in the control panel.

RainyDog
10th March 2016, 15:42
Sometimes my GTX 980 gets stuck in the wrong powerstate when opening videos, it doesn't run with full clock then and drops lots of frames. I suspect this is a randomly triggered bug by NNEDI3 in the OpenCL driver, a bit annoying.
Has someone else observed anything similar? Happens with the current driver (364.51) and some previous ones as well. Win 10 x64.

Add MPC-HC, or whichever video player you use, to 3D applications in the Nvidia control panel and change its power status to prefer maximum power.

aufkrawall
10th March 2016, 15:53
I don't think that this would help since the use of OpenCL/CUDA forces the card into a specific powerstate anyway.

leeperry
10th March 2016, 17:44
Enhance details: 1.0 is amazing for resolution
Kinda late to the party and I dunno what voodoo magic is behind "enhance details" but it really looks outstanding @0.7 with NEDI+2X SR@1+FRC+SSIM 2D 25% on SD :eek:

Werewolfy
10th March 2016, 17:45
Are you using DXVA deinterlacing or forced film mode? Does the problem occur with both? And which queue is the first one which is getting into trouble (less than half full)?


I use forced film mode. It seems that decoder queue and ivtc queue limit the other queues, it never fills completely. But the moment I have dropped frames is when render queue drops down.
I tested DXVA and it's not really better. Decoder queue go higher than before but I also have dropped frames when render queue drops down.

Here some videos of my tests, I hope it will be usefull and easier for you to understand.

http://www.filedropper.com/testdeinterlacing1_1

(It's the first time I'm using this file hosting website, I don't know what it worths).


Would be great if the new NVidia drivers improved the situation there, of course!!


Unfortunately, on my side it still doesn't work as intended even with the new Nvidia drivers, glitch handlling mode enabled or not.

Maybe try upping the GPU queue size even more?

I need to use a GPU queue of at least 20 to avoid dropped frames with interlaced NTSC DVDs. Any less gives me occasional and "unexpected" dropped frames. After extensive testing, 20-20-6 works for me perfectly.

(My setup is Win10, factory-overclocked GTX 960 and a dual core Ivy Bridge Pentium, using D3D11 windowed mode.)

Higher GPU queue doesn't help because it's limited by the decoder that's never filled. On my videos, you can see that I even used 128-24-15 without success.

Georgel
10th March 2016, 18:22
Kinda late to the party and I dunno what voodoo magic is behind "enhance details" but it really looks outstanding @0.7 with NEDI+2X SR@1+FRC+SSIM 2D 25% on SD :eek:

It does work nicely with real life videos.

It does not work well with animated videos, inducing grain, and what could be perceived as noise.

Uoppi
10th March 2016, 21:03
there are problem with deinterlancing with windows 10.

of cause this is not an madVR error.

I was just about to say Windows 10 + GTX 960 has worked quite nicely for me too.

Are you referring to some DXVA deinterlacing quality issues? At least deinterlacing seems to work glitch-free to my eyes, so I'm just curious what problems you are referring to.

iSeries
10th March 2016, 22:49
Only 2ms difference between 1D and 2D? I think something's broken, you should maybe look into resetting your settings as my GTX 960 has about 8ms render times when downscaling 4K content and 16-18ms when using 2D (with AR and LL)

It depends on how much downscaling is required but yes there is always some performance cost over 1d. However, specifically when supersampling, 2d adds 40ms to rendering times over 1d. Using 'Everest' as a test:

Example 1 (1080p supersampling):
Chroma - jinc
Doubling - superxbr 100
Enhancements - crispen edges 1, enhance detail 1, superres 3
Downscaling - ssim 1d
=30ms rendering time

Example 2 (1080p supersampling):
Chroma - jinc
Doubling - superxbr 100
Enhancements - crispen edges 1, enhance detail 1, superres 3
Downscaling - ssim 2d
=70ms rendering time

As you can see, the only difference between the two examples is 1d / 2d, with 2d adding 40ms to rendering times. Surely, that can't be right? Like I said, in every other downscaling or doubling and then downscaling scenario 2d is perfectly usable.

huhn
10th March 2016, 23:13
I was just about to say Windows 10 + GTX 960 has worked quite nicely for me too.

Are you referring to some DXVA deinterlacing quality issues? At least deinterlacing seems to work glitch-free to my eyes, so I'm just curious what problems you are referring to.

amd crimson deint is simply broken it is using a low quality bob. i haven't tested the newest beta yet.

and nvidia deint was kind of "broken" the last time i check and had issue with telecine/hybrid content. the high quality CUVID deint is not usable at all under windows 10.

huhn
10th March 2016, 23:14
I don't think that this would help since the use of OpenCL/CUDA forces the card into a specific powerstate anyway.

and that powerstate is the highes possible so it can help here.