I'll check the color banding when AUR is updated(Lazy to compile from git, sry). Good that it's clear that CMPlayer and VLC were in the wrong then, and of course nice job implementing dithering, i hope it'll be configurable (On/Off, Random/Ordered). But yeah it is strange that the colors looked better when they were wrong, hmm...
Now about "inactivated" i wasn't getting at that it was wrong because i had activated it, i was talking about linguistics, the word "inactivated" is very rarely used. So i was trying to say that it'd be more linguistically correct (more correct english) to say Disabled/Enabled instead of Inactivated/Activated, or just Inactive/Active, using "On" and "Off" would also be more correct. You could also use the word "Deactivated" which is much more commonly used than Inactivated and is basically an equivalent of Disabled, but i still think it'd probably be best to follow the trend of other programs (use Enabled/Disabled or On/Off) this might just be an opinion, and i'm no linguistics master, but i've just almost never seen the word "Inactivated" used, at least in a program thats specifying if a feature is on or off. And yes i deduced that you're probably Korean right? (Saw korean fonts on CMPlayers webpage) I've read some of boichi's mangas, Sun-Ken Rock is a favorite, but because of reading them i see all kinds of interesting things, foods in particular from Korea that i'd like to try, its a shame i live as far away as Iceland so it'd be extremely expensive for me to travel to Korea, Japan and China which are the places i'm most interested in visiting.
Anyways, back on topic, It's good to know that HWDec wasn't working because i was using a 10-bit encoded file, and putting "Unavailable" like you intend to for those cases would be helpful for those who encounter it. One other thing i'm curious about is why your player describes h264 as avc, most others would normally describe it as either h.264 or h.264/avc, since i didn't see h.264 anywhere i thought the input was being misread, until i finally remembered "ah, avc and h264 are the same thing", it also might be a good idea to change 420p into 4:2:0, many might mistake it for this, that is confuse the p for "Progressive Scan", they'd have to not really know what it is though since it links directly to the vertical resolution of the video to confuse it, but you can never predict humans.
I keep reporting my findings actively because i want this player to be good enough to be my player of choice on linux, since Mplayer/Mplayer2 are old, i've hated VLC for years both for experiencing lack of codec support, performance and quality within it, and Mpv has no GUI frontend, this is where i turned to next (closest thing to Mpv with a GUI frontend i could find) and to be honest i like it for the most part, you're doing a great job! but i still need it to match up to at least the video quality i can squeeze out of Mpv so that i will want to use it I have full faith that you'll eventually get that right.
Last edited by rabcor (2013-10-15 13:48:48)
@Kosava, and what font do you use? I think I have to test in same font configuration.
It is CP1250 encoding.
Fonts: Arial, Calibri, Liberation Sans, Sans Serif.
Fonts: Arial, Calibri, Liberation Sans, Sans Serif.
Same it show that letters artificed.
Dithering feature will be configurable in menu with three states(Off/Random Dithering/Ordered Dithering).
It's funny that you like the wrong color because, in fact, I've thought something's different with 10bit-encoded things and misunderstood like "Oh, 10bit encoded color is very vivid! 10bit is good!", too.
I'll fix the wrong expression. Thank you for letting me know that.
Since english is my second language(yes, I'm korean), I think there are many wrong or unnatural terms and expressions. I need many feedbacks.
And... yeah, iceland is too far from here. I've been in Japan for about 6 years and I agree with you. Japan is interesting country, and of course I really love korean food.
You can see the H.264 in Video Codec item (H.264 / AVC / blah blah...). the strings avc1 and 420p are handed over from mpv side.
It's easy to remap those strings to another ones in CMPlayer but I doubt its usefulness.
CMPlayer is not an encoder or decoder which is sensitive those informations but a player which just handles given formats.
The playback informations are just hints to confirm what's going on or that things are applied properly.
If someone wants to know those things, he or she is capable of noticing what means what by himself or herself like you usually.
Although I agree that the 420p can be misunderstood as 420 progressive resolution, in most case, it's harmless because it makes nothing changed.
So if I have to make some changes, I'll take more specific descriptions like "4:2:0 planar (10-bit per component)" but it can look too long.
Greatly thank you for all your contributions.
@Kosava Can you send me the subtitle file, too?
Last edited by xylosper (2013-10-16 02:13:38)
Thank you. There's a bug related with encoding selection. First of all, the right encoding is not CP1250 or UTF-8 but CP1252.
The problem is CP1252 is not selectable which is related with autodection.
I'll fix it and the change will be included in next release.
Last edited by xylosper (2013-10-21 03:07:38)
0.8.0 is released. For details, see the release page: https://github.com/xylosper/cmplayer/re … tag/v0.8.0
@Kosava Now you can select CP1252 in subtitle encodings. Btw, I have tested your file and the autodetected encoding was wrong. So you have to disable autodetection or raise the accuracy to enfoce autodetected encoding(when I tested, 80 was fine).
@rabcor Some interpolators are added. I think 3-Lobed interpolators(spline or lanczos) produce quite sharp pictures. Please try them. Note that '3-Lobed Lanczos' is different from '3-Lobed Lanczos (Approx)'.
I really like the changes you made to the playback information, much better! Dithering options are great to have and the colors now roughly match that of Mpv. I really like how many options the players has by now, i love options!
I'll give you a little report of the issues i've found so far:
i'm using the same file as before at the same scene for testing, i set interpolator & chroma to lanzcos 3.
The biggest problem is that i see a tiny green border on the right side of my video. This applies to both windowed and fullscreen mode. I think this only happens with 10-bit files.
Upscaling still seems to be nowhere near as good as in Mpv, but that might be related to the sharpness issue (below)
The sharpness of videos has not been fixed, i can confirm however that it is not an upscaling issue (comparing original resolution file in windowed i can still see this difference in sharpness) my point of reference in this is the samurai's face in the picture, what i'm talking about is actually visible in the old screenshots. I also notice some "Color Contagion" which may be related. What i mean is that in the images i took before, there are letters: Roadshow which have a yellow outer glow effect, this effect seems to affect a bigger area in CMPlayer than in Mpv, making an area that is grey in mpv grey with a small green tint. It might be something else than color contagion though. For example i notice that the white pages in the pictures seem more white in Mpv than in CMPlayer.
If i enable the "Sharpen" filter which creates an ugly & extreme sharpness effects, that "color contagion"-like effect disappears and the image is much sharper in CMPlayer than in Mpv, but its just too sharp so it turns ugly. (So in other words it might be a sharpness setting of some sort that's causing this)
I tried changing the "Video Filter" settings in the preference menu, it seems to have little/no effect.
"Invert Color" & "Grayscale" filters, at least in this 10-bit file turns everything green rather than inverting the colors/turning greyscale.
Changing a workspace (openbox, mousewheel) stops playback if in windowed mode.
Exiting fullscreen via double click (sometimes) minimizes window (openbox) <- i think a reproduction method is to open a video, put it to fullscreen, switch to another workspace then back to the one that has the video and double clicking to go back to windowed. The minimization will also stop playback if it's playing.
Changing Color Range in the same 10-bit encoded file, the default (auto/mpeg/limited range) shows the same colors as Mpv as it should, the JPEG/full range setting seems to put a grey overlay over everything rather than changing the colors to the way they were with incorrect color range before the patch. Are you sure you did this right?
Suggestion: In the settings menu for Deinterlacing it says "Make framerate doubled" i'm pretty sure it would be more correct to just say "Double framerate"
Suggestion: Launching a video in windowed mode doesn't set the window's size to the videos resolution, rather up/downscales it to the last window size. This can be done in the player by pressing "1", but i think it'd be best if it was a default setting so that everytime you open a video in windowed mode it will automatically resize the window to the video's size. I think most other media players do that.
Last edited by rabcor (2013-10-23 20:16:17)
All related bugs with openbox which @rabcor reported in previus post, i can confirm, I'm use openbox too.
As a matter of fact, I cannot see the difference in sharpness in your screenshot and also in my desktop. Maybe my eyes are too insensitive. However, I found a bug related with the green line which is caused by wrong chroma location. Maybe this affects what you called 'Color contagion', I think, and also sharpness. I fixed it so please try it again in next release. Also, Invert Color and Grayscale filter is fixed now.
You can customize the sharpness by sharpen filter but you need some knowldgement about kernel filter for image. Related options are in 'Video filter' tab in preferences.
And, yes, I'm sure that full color range shows gray-overlaid image because white in limited range(240) is no more white in full range(255).
Maybe what you want is remapping the range linearly. I can add another option for that but I think the result is merely different with original limited one.
I want to test openbox, I'm not familar with it. I installed and launched openbox but I couldn't see any of workspace related options.
Anyway, for now, you can disable the 'auto-pausing' behaviour in playback tab in preferences.
Your eyes can't be insensitive to what i'm seeing, the difference is that big! i'll take other screenshots in a bit. So you can see what i'm seeing
Read the arch wiki page for openbox configuration, a raw openbox setup with no additions (such as a taskbar or a clock) only has right click and mousewheel for two workspaces if i recall correctly.
You need i think 2 programs for configuring it. obconf & obmenu. an automated alternative to obmenu (so the right click menu will be automatically generated) is menumaker. These are all available in the official repositories.
I like to run menumaker first, then obmenu to customize it afterwards. Obconf is a must have. Beyond this i haven't really customized my openbox yet, but i can see a lot of customizing potential in it since it starts you off with nothing and you can add whatever you want to it. It's so much the "arch way". LXDE known for being the most lightweight desktop environment uses Openbox as a base, so maybe you can get the bugs we get on openbox in LXDE.
Once i get the hang of Openbox though, i'm gonna try Awesome and Xmonad again and eventually decide which of the 3 i like best.
Last edited by rabcor (2013-10-24 15:47:56)
Thank you for your instructions. I'll try openbox again.
FYI, please use png not jpg for screenshot. Since jpg uses lossy compression algorithms, quality regression always occurs in jpg format.
One more bug noticed, in windowed mode when i want to load subititle when i click on right click to open subtitle, window for opening subtitle apear behind video window.
@Kosava Thank you. I fixed it.
Two bugs are fixed: minimized when exit fullscreen mode and subtitle dialog issue. These are included in next release.
Another issue 'Changing a workspace (openbox, mousewheel) stops playback if in windowed mode' is not fixed and I think it's a bug of Qt or openbox.
CMPlayer's auto pausing works when the window is minimized and I figured out that Qt always reports that the window is minimized when the window is outside of current workspace in openbox.
Since this is a bug related with framework, I cannot do anything and this should be fixed in Qt or openbox(I'm not sure which one is wrong).
Anyway, if you think this is annyoing, the only workaround which is available currently is to disable the auto pausing option.
About the sugguestion to fitting the window size to video resoultion, in fact, I hate the behaviour and I always turn it off.
This cannot be an issue if serial sources have the same resolution.
With difference resolution of sources, every resizing the window makes other jobs disturbed in several forms, e.g., covering other window.
And, if the video resolution is larger than the monitor resoltion, the result cannot be defined, escpecially, in Linux because oversized windows could be shown differently for each WM or DE.
I took these images with gimp, and exported them as uncompressed PNG files, just the way you want them
I finally got around to taking these screenshots Turns out, the sharpness difference i was talking about seems to now be barely visible, it might have had to do with restarting something (either way i've restarted my computer inbetween, last time i was running it right after getting the package installed from AUR) visible though, but mostly the color difference i'm talking about is clearly visible in the pictures below. Remember my point of reference in both examples is the samurai's face and the book cover itself (if you want to see it like i do). Another good point for reference is the button on the girl's shirt.
This time i also took fullscreen samples and now a sharpness difference can clearly be seen, but this time of course is due to upscaling, my shitty connection wanted to murder me while i uploaded these two.
Remember to zoom them up to actual size
Ignore the black box in the upper left corner, i think that's an openbox bug it's probably.
Settings below are the only modified ones after default settings.
hwdec = vdpau
Interpolator & Chroma Upscaler: 3-Lobed Lanzcos
Video Filter (Preferences): All set to 0
Deinterlace S/W & H/W: Bob, Double Framerate, OpenGL, HwAcc(H/W) (set to Auto for the video)
Hardware Acceleration: All On (Disabled because of 10-bit)
There are differences between both players while in windowed mode, these differences become much more clear after upscaling is done.
Last edited by rabcor (2013-10-25 18:41:52)
Great! Though I still cannot notice the difference when unscaled, I can see the difference in fullscreen and I fixed something.
How about this one? This is taken from next release candidate.
0.8.1 is released. This version contains minor patches and a new function.
@rabcor I added new Color Range 'Remap Range' and 'Extended Range'.
Remap Range rempas the range for luma (16..235 -> 0..255) and chroma (16..240 -> 0..255).
Extended Range remaps both of luma and chroma to same range of luma. This corresponds to extend the range for chroma and clipping the overshooting. The result is definetely distorted but close to 'the previous wrong color', I think. Please try it.
What resolution screen do u have? you should easily be able to see it, i'm on a 1920x1080 screen myself.
I'll check the color ranges later
I checked your screenshot, it's much much more presentable than before! i actually like it. Your shot is much sharper than the Mpv shot i took! what exactly did you do?
Looking close though, the (quite minimal) color difference is still there, it did not change which means we can assume it probably doesn't relate to sharpness. A good point of reference is the middle part of the book, the colors of the cover around the Japanese characters show this difference pretty well.
Another thing i notice is something that looks like an artifact, again something that needs to be looked very close at to see. There's a yellow "X" right next to the samurai, above the left end of it is where i see it. It looks not dissimilar to color banding and might have to do with the upscaling, I'll have to check it on my end to further investigate i'm gonna add a k-lite/madVR screenshot into the mix too.
One more bug noticed.
I use pcmanfm and when i want to open video file from pcmanfm file manager, cmplayer write this error
Cannot open the media.
Also i tried qtfm (other file manager) they not write error output but not start video too.
Double Commander Also not work same bug like for pcmanfm.
Edit: Thunar, Nemo, Nuatilus, Dolphin work fine.
Last edited by Kosava (2013-10-27 11:01:48)
@rabcor My monitor's resolution is 1920x1080, too. Honestly, I have no idea why the color is different. I've compared every mathematics but I cannot find anything wrong for now.
@Kosava For pcmanfm, it'll be fixed in next release. However, for qtfm, I think it's the problem of qtfm. qtfm calls cmplayer with wrong arguments. cmplayer should be called as described in .desktop file, 'cmplayer --wake-up --open %u' but qtfm calls 'cmplayer "--wake-up --open %u"' and this is not a valid option for cmplayer, of course.
Last edited by xylosper (2013-10-27 16:17:13)
I just discovered CMPlayer and I must say it is excellent! Thank You and keep up the good work
@cappuccino Thank you!
0.8.2 is released. This version provides some minor patches only. For details, see https://github.com/xylosper/cmplayer/re … tag/v0.8.2
I was just about to take those screenshots i had promised but as it turns out the current AUR version doesn't seem to work, i also notice an older git version is in AUR now whats up with that?
../../build/lib/libcmplayer_mpv.a(resolve_quvi.o): In function `mp_resolve_quvi': resolve_quvi.c:(.text+0xc4): undefined reference to `quvi_init' resolve_quvi.c:(.text+0xf9): undefined reference to `quvi_supported' resolve_quvi.c:(.text+0x13b): undefined reference to `quvi_setopt' resolve_quvi.c:(.text+0x14d): undefined reference to `quvi_parse' resolve_quvi.c:(.text+0x18f): undefined reference to `quvi_getprop' resolve_quvi.c:(.text+0x1ad): undefined reference to `quvi_getprop' resolve_quvi.c:(.text+0x1bf): undefined reference to `quvi_parse_close' resolve_quvi.c:(.text+0x1c9): undefined reference to `quvi_close' resolve_quvi.c:(.text+0x2d8): undefined reference to `quvi_strerror' resolve_quvi.c:(.text+0x2fb): undefined reference to `quvi_close' collect2: error: ld returned 1 exit status Makefile.Release:361: recipe for target '../../build/cmplayer' failed make: *** [../../build/cmplayer] Error 1 make: Leaving directory '/tmp/yaourt-tmp-illustrious/aur-cmplayer/src/cmplayer-0.8.2/src/cmplayer' Makefile:34: recipe for target 'release' failed make: *** [release] Error 2 make: Leaving directory '/tmp/yaourt-tmp-illustrious/aur-cmplayer/src/cmplayer-0.8.2/src/cmplayer' Makefile:36: recipe for target 'cmplayer' failed make: *** [cmplayer] Error 2
Last edited by rabcor (2013-10-28 12:42:37)
I can confirm not work new version dont want to compile...
Added some patch. Please try again.
@rabcor I'm not the maintainer of git package. Anyone can make and upload a new package which he or she want. In fact, the maintainer have asked me about git package and I refused: https://github.com/xylosper/cmplayer/pull/8
Last edited by xylosper (2013-10-28 21:19:35)