Guest - Testers are needed for the reworked CDateTime core component. See... https://forum.kodi.tv/showthread.php?tid=378981 (September 29) x
  • 1
  • 56
  • 57
  • 58(current)
  • 59
  • 60
  • 532
Kodi DSPlayer – DirectShow Player for Windows
What Is HDR Tone Mapping?

Tone mapping is way to remap color volume from a display with a larger color volume to a display with a smaller color volume. This is done when a display lacks the same peak luminance as the original source Maximum Content Light Level (MaxCLL) and needs to tone map those brightness levels that are too bright for the display to represent. Tone mapping also involves saturation mapping that remaps color saturation as the color gamut is compressed by tone mapping to ensure all compressed colors fit within the visible display color gamut. Combined, these two processes are referred to as tone mapping and gamut mapping.

As of 2019, tone mapping of HDR10 content very much remains the Wild West. Every display manufacturer is free to choose how to represent HDR content on its displays. This has led to varying tone mapping approaches with varying results. Many consumer HDR displays are still unable to reproduce the full range of light output necessary to match HDR mastering monitors and display manufacturers have come up with different compromises to make the most out of each display type's strengths and weaknesses.

Current HDR display types:

Backlit LED TV: Calibrated peak brightness range of 300-4,000 nits.
Edge-Lit LED TV: Calibrated peak brightness range of 300-2200 nits.
OLED TV: Calibrated peak brightness range of 540-800 nits.
HDR Projector: Calibrated peak brightness range of 50-250 nits.

HDR videos are mastered to different levels of peak brightness. The most common video peaks are 1,000 nits, 4,000 nits and 10,000 nits. These source peaks are communicated to the display through static HDR10 metadata embedded in the source file: MaxCLL (Maximum Content Light Level), MaxFALL (Maximum Frame Average Light Level) & mastering display maximum luminance. This metadata informs the display of the video's maximum pixel brightness and provides it with the option to adjust its tone curve to deal with the source.

In most cases, the display will select a different tone curve for 1,000 nits, 4,000 nits or 10,000 nits sources that is suitable for representing the dynamic range of the source. Sources with a higher reported frame peak brightness have a higher dynamic range and generally require more aggressive tone curves to compress the source levels to fit within the display range. Other HDR displays simply ignore the HDR metadata and tone map every video the same way.

Three types of tone mapping curves are used by current HDR displays:

Example of Tone Mapping Curves (Batman v Superman): Soft Clip vs. Highlight Roll-off vs. Contrast Curve

Soft Clip 
The display follows the PQ EOTF curve 1:1 (or as close as possible) almost up to the display peak white and uses a sharp roll-off curve at the top to clip any specular highlight detail beyond this limit.

Soft clipping maximizes use of the display's peak brightness at the expense of losing detail in the brightest specular highlights. This approach maintains the accuracy of the original source APL (Average Picture Level) as much as possible while ignoring any lost detail in the brightest parts of the image.

The screenshot linked above is from the “Dream Sequence” in Batman v Superman: Dawn of Justice. This sequence displays blinding specular highlight detail in Ben Affleck’s white shirt punctuated by a flash of lightning in the background with a frame peak of 4,865 nits. Dimmer displays that start the roll-off point late in the display luminance curve tend to lose most of the detail in Ben Affleck’s shirt and the lightning bolt, turning them into white patches without visible detail. 

Sample Tone Curve Shape
Real-World Equivalent (Sony A1 OLED)

Highlight Roll-Off
The display rolls-off all or some of the specular highlights above SDR reference white (0-100 nits) with a more gradual tone curve that most often starts within the last third of the display's luminance curve to compress any source levels that are beyond the display peak limit.

The highlight roll-off is superior at preserving bright specular highlight detail with less clipping at the expense of dimming all source values above the roll-off point. Highlight roll-offs also preserve most of the APL of the image by avoiding tone mapping of reference white and limiting compression to the specular highlights.

In the same 4,000 nits scene from Batman v Superman: Dawn of Justice, more of the detail in Ben Affleck’s white shirt and the lightning bolt would be visible, but depending on where the roll-off point is started and its chosen shape, not all highlight detail may be retained.

Sample Tone Curve Shape
Real-World Equivalent (Panasonic EZ1002 OLED)

Contrast Curve 
The display starts a gradual, arching tone curve lower on the PQ EOTF that starts early enough that some of SDR reference white (0-100 nits) is compressed and most of the source range is compressed.

The contrast curve is best for replicating the source's high contrast at a lower peak brightness at the expense of reducing the brightness of most source values uniformly across the display range. A contrast curve is used to create a smoother transition from the midtones to the highlights by lowering the midtones to enhance the contrast between the specular highlights and reference white.

In the same 4,000 nits scene from Batman v Superman: Dawn of Justice, the display would dim the overall image slightly, but all of the detail in Ben Affleck’s white shirt and the lightning bolt would be retained with greater contrast and separation between the specular highlights and surrounding background.

Sample Tone Curve Shape
Real-World Equivalent (LG C7 OLED)

By this definition, a contrast curve could be considered the only "true" tone mapping curve. It puts the greatest emphasis on reproducing the original appearance of the HDR source by mirroring the source's higher dynamic range and preserving all specular highlight detail within the display's more limited dynamic range. However, in the interest of displaying a consistently bright and pleasing image, many HDR displays will sacrifice some of the source's high contrast to maintain consistently higher APLs (Average Picture Levels).

Unfortunately, the tone mapping approach used by each display is hidden to the viewer and can often only be revealed by taking measurements of the screen with appropriate test patterns and a colorimeter. About the only thing that can be inferred is that an HDR display such as a projector is tone mapping more of the source range than a 700 nit or 2,000 nit flat panel TV.

HDR Display Type Strengths and Weaknesses

Different display technologies (e.g., Backlit LEDs, Emissive OLEDs and Front Projectors) have different strengths and weaknesses in representing HDR content that are inherent to its technology and the amount of light output it can create under different circumstances. This can make different display types great at representing certain types of HDR scenes, but poor at others.

Backlit LED TVs, due to its high peak brightness, are strongest at representing high APL scenes with many bright image areas. They can push the whole screen to be very bright (have the ability to render scenes with a high MaxFALL), yet struggle to light up smaller bright areas isolated against darker backgrounds. LED displays usually have a limited number of local dimming zones that are forced to simultaneously dim dark portions of the screen while boosting the bright portions. The larger dark areas cause any nearby smaller bright areas to be lowered in brightness to produce necessary contrast, often with some light haloing visible around the edges of the bright objects. As a result, most LED TVs cannot reach its stated peak brightness when rendering the many small highlights found in HDR videos such as smaller light sources in dimly-lit image areas due to the reliance on local dimming to create bright/dark contrast.

Emissive OLED displays are heavily limited by ABL (Auto Brightness Limiting), so scenes with full-field brightness are dimmed aggressively by the display (the display has a limited ability to present scenes with a high MaxFALL). But because the display can turn each pixel on and off independently, OLEDs are very effective at representing small and bright specular highlights against darker backgrounds that can provide them with an advantage compared to LEDs in reproducing the local contrast intended for the majority of HDR specular highlights found in HDR videos, as well as the ability to make the most out its stated peak brightness in these scenes. The strengths of OLEDs are aided by the plight of many current HDR mastering monitors used to create HDR content that are also highly limited in full-field brightness due to ABL (Update: Brighter LED-based mastering monitors are already on the way).

HDR front projectors are severely limited in light output compared to TVs. As a result, they don’t stand out in any particular category when displaying HDR content, but the best projectors have greatly benefited from advances in tone mapping that have made low peak brightness HDR, as low as 100 nits peak, have improved visual impact over legacy SDR content. Tone mapping PQ HDR to SDR brightness levels can result in sharper images with more defined contrast over simply mastering the source video in 100 nits SDR gamma.

So how is a tone mapping curve constructed and how is it applied?

Tone Curve vs. Tone Mapping Curve

In order to tone map, you need a tone curve.

A tone curve is the basic foundation for representing any kind of image, print, digital or video. Consumer video displays use a range of internal tone curves described by terms such as gamma 2.20, gamma 2.40, HLG, Dolby Vision or HDR10. What tone curves do is distribute image tones across the available contrast range. A tone curve works equally on all three RGB color channels — so it impacts the global brightness or luminance channels of the image. The brightness range of any display is limited, so each tone curve prioritizes the available tonal range by emphasizing certain image tones. In image processing, these image tones are subdivided into three unique boundaries:
 
  • Shadows: The darkest parts of the image near black;
  • Midtones: The middle range where the majority of the image is located;
  • Highlights: Brightness peaks distinct from the midtones located near peak white.

Three tonal ranges of an image:
Image

In most image grading software, the image tones are represented in a curves panel bisected by a straight line that stretches from the bottom-left corner to the top-right corner, with the shadows represented at the bottom of the panel and the highlights at the top. The display's full range of contrast is restricted within these boundaries: so adding or subtracting from any group of tones requires removing or adding from another; you can't add more brightness to the highlights without losing some of the midtones or shadows. Therefore, tone curves do not create any new tones; they merely redistribute contrast from one area to another.

Because HDR video prioritizes images with a wider range of contrast, most color grading for HDR content is focused on tone curves that emphasize brighter highlights and contrast enhancement. This is often achieved by use of an S-curve.

S-curves can be applied to the tonal range of any scene by simply pulling down the shadows at the bottom of curves panel to create deeper blacks and pulling up the highlights at the top to create brighter highlights. This produces an S-shaped arc that etches the shadows and calls attention to the highlights, which has the combined effect of creating images with higher contrast.

Curves Panel in Photoshop

S-curve tone curve:
Image

Tone mapping is concerned with reproducing images with a range of contrast that is greater than the display can represent. When contrast is compressed by an S-curve, the straight middle section of the curve is used to preserve midtone contrast with the brunt of tonal compression being applied to the shadows and highlights.

The shape of an S-curve is meant to mimic the response of film stock. The PQ SMPTE standard (ST.2084) applies an S-curve in its OOTF (Optical to Optical Transfer Function) to convert scene or camera light into display light. Accurate replication of HDR video for consumer displays then may use an S-curve to preserve the same intended look and feel of the original HDR footage.

What Is a Tone Mapping Curve?

Tone mapping for HDR video involves two steps: tone mapping and gamut mapping. The two processes are used to compresses the dynamic range of a video with combined color volume mapping to map a larger color volume to a smaller color volume.

A tone curve is used to compress the peak luminance of the source so that the brightest pixels are mapped to the precise peak white luminance of the end display. Classic tone mapping curves use an S-shape that tilts away from the PQ curve to roll-off the brightest source values up to the display peak. The curve's S-shape maintains relative relationships between the shadows, midtones and highlights compared to the HDR source reference.

The basic make-up of a tone mapping curve includes a soft toe at the bottom that sets the lower contrast point, a linear section intended to preserve midtone contrast and a steep shoulder that rolls-off any highlight information.

Anatomy of a tone mapping curve:
Image

The trend among current HDR TVs (especially bright LEDs and OLEDs) is to forgo a sweeping S-curve for most content in favor of a simpler highlight roll-off (particularly for the majority of HDR content that is being mastered to 1,000 nits).

The highlight roll-off has several advantages for true HDR displays with a stated peak brightness of 540 nits or greater. Limiting tone mapping to only the specular highlights is a response to the majority of viewers' preferences for higher APLs over retaining every last bit of specular highlight detail in the source. A highlight roll-off does not tone map the midtones, which leads to more consistent APLs that are closer to how the video would appear on a mastering monitor. And focusing on rolling-off the specular highlights in isolation reflects the reality that 90% (or more) of HDR video levels are currently mastered in the 0-100 nits SDR range. Most displays have more than adequate brightness to render 0-100 nits (SDR reference white) without compression, with some headroom still remaining to roll-off the small number of brighter source values.

Highlight roll-offs may remove the toe section of the curve and instead use a linear curve with a shoulder section that is adjusted in size to roll-off any highlight information. Content mastered to 4,000 nits or 10,000 nits may deepen the roll-off and add a toe section, but some type of basic highlight roll-off could also be used.

Projectors are the only HDR displays known to consistently make use of a full S-curve to display HDR videos to reflect a projector’s especially limited dynamic range compared to the HDR reference mastering monitors used to master HDR content.

Gamut (or Saturation) Mapping

Tone curves are an effective method of redistributing luminance from one display to another. However, gamut mapping is needed to address the other component of color: chrominance.

Gamut mapping is a way of mapping color saturation from one gamut volume to another. Any display with lesser color gamut coverage than a mastering monitor (which is currently DCI-P3) and all displays that employ heavy luminance compression require saturation adjustments after tone mapping to adapt to differences in gamut shape and volume.

HDR displays are asked to reproduce a wider range of colors that can be both very bright and highly-saturated. Large decreases in peak brightness caused by tone mapping also create significant reductions in the amount of available color saturation, particularly for colors along the edges of the color gamut such as the bright reds, yellows and greens used by HDR videos. So, as gamut volume is lost, some highly-saturated colors along the edges of the gamut will become too bright and saturated for the display to represent with its more limited color volume. These highly-saturated colors are addressed through a desaturation step after luminance compression to match any out-of-gamut colors with an in-gamut value that the display can represent within its limited peak brightness.

What Is Gamut Mapping?

Complete tone mapping can then be summarized by a combination of a tone curve that compresses luminance (reduces the source dynamic range) and saturation mapping that desaturates colors to fit within a reduced color volume (preserves relative color luminosity).

Examples of Tone Mapping with DaVinci Resolve

A visual way to represent tone mapping curves is to represent HDR videos in waveform graph. A waveform graph illustrates the distribution of pixel brightness of all pixels in a given video frame by plotting the peaks and valleys in luminance present within the frame. This creates a continuous wave shape that provides a visual representation of the brightness and positioning of the shadows, midtones and highlights in the image. HDR waveforms can be used to show the before and after effect of applying a tone curve to a video by observing changes to the distribution of pixel luminance after tone mapping is applied. 

Popular color grading software DaVinci Resolve will be utilized to produce some sample HDR waveforms. Davinci Resolve is used frequently to color grade some of the HDR video content available today (Resolve is available in both free and paid Studio versions).

The sample scene is from The Greatest Showman UHD Blu-ray with a frame peak of 1,197 nits. When imported into the Resolve Timeline editor, HDR videos are displayed in an HDR format, so the screen capture below appears somewhat washed out when viewed on an SDR monitor.

The Greatest Showman in DaVinci Resolve (1,197 peak nits):
Image

Waveforms in Resolve are presented in the scopes panel that displays every pixel in the frame separated into its red, green and blue color channels and plotted relative to its brightness and position in the image.

Resolve also contains built-in HDR controls to apply both tone and gamut mapping to the video. With these controls enabled, the scopes panel shows the impact tone mapping has on the distributed brightness of the frame.

The full waveform for the frame above is shown below. The graph is labelled with the peak display nits corresponding to some of the notable 10-bit PQ values indicated on the graph.

DaVinci Resolve HDR Scopes Peak Luminance Waveform:
Image

As you can see in the graph, the frame does indeed peak at 1,197 nits, where the two rings of fire are located. Now Resolve's tone and gamut mapping is applied to compress the peak brightness from 1,197 nits (typical LED brightness) to 700 nits (typical OLED brightness).

The result of applying the tone curve is shown below. At 700 peak nits, tone mapping hardly alters the distribution of the pixels in the original image. Most compression by Resolve has been applied to the small stacks representing the specular highlights (with some partial clipping of the rings of fire that is difficult to see in the screenshot). Enough contrast exists between the 700 nits display peak and reference white to properly present the intended contrast of the scene, so the roll-off limits compression to the specular highlights and leaves the bulk of the image below 100 nits undisturbed. 

Tone Mapping Curve - 700 display nits (HDR):
Image

Things get more interesting when the available display peak is restricted more severely. Compressing the same frame to 100 nits (typical HDR projector brightness) necessitates more compression to the base of the image to replicate the scene's higher contrast.

In this example, the image is converted from HDR to SDR by madVR at 287 target nits and captured at 100 display nits to look correct on an SDR monitor. You can see in the screenshot below that after gamma conversion the color is returned to the image and the entire scene is brought back to life.

Some precision is lost in this case because the input into Resolve is a BT.709 lossless screenshot from madVR and the output is ST.2084 PQ values. However, comparing screenshots to 1:1 output from the original source footage in Resolve shows the waveforms produced by madVR are actually surprisingly close in pixel luminance and can be considered fair representations of the HDR to SDR conversion.

Even at 100 peak nits, the highlights again receive the majority of the compression. However, to represent the bright highlight detail above 100 nits, some headroom has to be created by reducing 0-100 nits to a lower brightness. This slightly lowers the image APL but is necessary to retain the higher contrast of the scene with the same relative waveform shape.

Tone Mapping Curve - 100 display nits (HDR to SDR):
Image

The screenshot below summarizes the levels of the compression from 1,197 nits, to 700 nits down to 100 nits. In each case, the waveform shape remains quite similar with tone mapping rescaling the contrast of the scene to fit the peak luminance of each display.

The tall peaks receive most of the tone compression. The 700 nit and 100 nit waveforms differ more in APL from the more severe limitations on brightness imposed at 100 nits. The difference in APL between the tone mapped image at 100 nits and original image at 1,197 nits could cause the image to appear overly dim on a 100 nit display, yet the compression created by tone mapping still works because our eyes are more sensitive to changes in contrast than changes in peak brightness. The lower APL appears visibly brighter than it otherwise would because the original scene’s contrast (the difference between bright and dark image elements) is largely retained through tone mapping.

Original peak vs. 700 display nits vs. 100 display nits:
Image

Here is a second example, this time from the LG Cymatic Jazz HDR Demo. This scene is also captured in HDR from Resolve, so the color space is not correct, but the image is sharp enough to show the bright detail of the lightbulbs.

LG Cymatic Jazz in DaVinci Resolve (10,000 peak nits):
Image

The waveform shows the frame peaks at a very bright 10,000 nits.

LG Cymatic Jazz Waveform:
Image

This time, rather than use a classic S-shape tone curve, a simple highlight roll-off is used. To illustrate the difficulty in representing this especially bright scene, the roll-off is applied with the intent of keeping all specular highlight detail within a mere 100 display nits.

Highlight Roll-Off - 100 display nits (HDR):
Image

As you can see, the image is basically destroyed due to severe clipping and any attempt by the highlight roll-off to preserve the creative intent of the scene has failed.

It is often a challenge to determine if an HDR display uses any form of roll-off curve that clips very bright highlight detail due to the often inconsistent use of highlight peaks in current HDR videos. One approach to test this is to use white clipping patterns mastered to 1,000 nits, 4,000 nits and 10,000 nits to see how the display handles static HDR metadata mastered to different levels of peak brightness and look for any loss of visible steps at the ends of the patterns.

Others go-to tests for tone mapping include demo-worthy scenes in HDR movies with frequent and obvious use of very bright specular highlights. One of the most popular tone mapping torture tests is the “Sandstorm Scene” in Mad Max: Fury Road (00:25:18 Chapter 3) that features frequent bursts of specular highlights mastered as high as 10,000 nits and bright colors that can appear undersaturated when gamut mapping is improperly applied by the display

Mad Max Fury Road Sandstorm Scene:
Clipping and Improper Gamut Mapping 
Correct Tone and Gamut Mapping

In the final example, the same frame with the lightbulbs is again compressed, but this time with a more aggressive S-curve as opposed to a simple roll-off curve. Resolve's HDR to SDR conversion is applied at 500 target nits. The result of this curve is a presentation of the original scene that better reflects the display’s more limited peak brightness.

Tone Mapping Curve - 100 display nits (HDR to SDR):
Image

The tone curve applied by Resolve is not perfect, as it does clip some detail in the lightbulbs. However, the image looks much more representative of the reference frame.

This example also makes it easier to see the use of gamut mapping. The original highly-saturated yellow becomes desaturated to more of a wine color after tone mapping. This saturation mapping is necessary due to the target display's lower color volume at 100 nits that necessitates desaturation of the once bright yellow hue so it appears more natural and less oversaturated at this low luminance.

In summary, tone mapping is a reliable method of remapping the peak light output of any HDR video to the capabilities of lesser displays without destroying the creative intent of the source content or having to accept a dim or clipped HDR presentation on any type of display.

Further Reading:
Video: HDR Tone Mapping Explained
Should HDR Displays Follow the PQ Curve?
Determining HDR Display Performance with Spears & Munsil UHD HDR Benchmark Disc
Reply
@mkanet: if your internet streams work with the merits setting and not with mediasconfig/filterskonfig then there is a problem with you mediasconfig/filtersconfig. this has nothing to do with win 8.1. i also use win 8.1. please test with my mediasconfig and make sure there are all these thinks like lavsplitter... defined in filtersconfig.
Reply
@MKANET
Sorry, my question was only focused on achieving 1080p from Kodi playing Youtube (only players DASH codec compatible, afaik, can), instead of being limited to 720p maximum. Apart from that, YouTube works on my DSPlayer. Didn't want to mess up into the topic u were already discussing Wink
HTPC
Silverstone Grandia GD05 - Intel i5 3570k -Asus H61M-G Micro-ATX - Unidad Blu-ray
MSI GTX970 4GB GDDR5 - 8 GB RAM DDR3 - AVR Denon X3400H Atmos - LG  OLED 55C7V
Reply
(2015-07-06, 13:20)Bjur Wrote:
(2015-07-05, 23:04)Warner306 Wrote:
(2015-07-05, 01:00)Bjur Wrote: I have a problem with rc1 dsplayer madvr. Often when I stop a video i only see an empty frame so I have to exit to get normal screen. Any fix for that?

Did you try enabling Force madVR to exit fullscreen before stop in DSPlayer settings?

Or do you mean pause instead of stop?
Yes I did that. I have enabled madvr exclusive mode and kodi in window mode. Is it correct

Yes, that sounds fine. But you need to explain your problem in more detail.
Reply
Sorry will do.
My setup is that I have 2 machines which is sharing MySQL between them.
I have setup LAV filters and madvr and it has worked earlier 14.2. Then I deleted completely to start from the beginning with v. 15 RC1.
I now have problems when I play a video and press stopp to return to home menu. Many times I only get a page, where i can see the background of the skin. In the top left corner of the screen I can see text TV shows, series name etc. but thats it. I have no other text with episodes pictures etc. and I have no idea what the problem can be. I have tried confluence and AEON Nox skin, the same problem on both and when I restarts Kodi I get normal menus until next time.
Reply
(2015-07-07, 08:35)Bjur Wrote: Sorry will do.
My setup is that I have 2 machines which is sharing MySQL between them.
I have setup LAV filters and madvr and it has worked earlier 14.2. Then I deleted completely to start from the beginning with v. 15 RC1.
I now have problems when I play a video and press stopp to return to home menu. Many times I only get a page, where i can see the background of the skin. In the top left corner of the screen I can see text TV shows, series name etc. but thats it. I have no other text with episodes pictures etc. and I have no idea what the problem can be. I have tried confluence and AEON Nox skin, the same problem on both and when I restarts Kodi I get normal menus until next time.

You will have to provide a debug log by following the instructions on the first page of this forum. Did you also update your copy of madVR?
Reply
No I did not update it
Reply
(2015-07-07, 17:00)Bjur Wrote: No I did not update it

I'd update madVR first.
Reply
(2015-07-05, 01:00)Bjur Wrote: I have a problem with rc1 dsplayer madvr. Often when I stop a video i only see an empty frame so I have to exit to get normal screen. Any fix for that?

I have the same problem, and blank screen once I close a video, sometimes if I wait a few seconds it'll come back up, sometimes it won't.
Reply
Things appear fine on my end. I'm using the Direct3D 11 Presentation Path under General Settings. It must have something to do with changes to madVR if this is new.

Debug logs would be necessary, I would assume.

Someone mentioned the possibility madVR is not closing correctly with recent releases after stop. I have noticed some bad lag in the menu after stopping a video. I haven't attempted to debug this.

Again, a debug log would help.
Reply
When updating madvr should you uninstall first or just overwrite files? EDIT: I uninstalled and overwrite old files with new files and installed again. Will report back if the problem is still there.
question 2: I have defined that sd contest should use xxx madvr, and 720p yyy etc. but it is all defined in Kodi and not in madvr. Is this okay?
Reply
(2015-07-08, 13:43)Bjur Wrote: When updating madvr should you uninstall first or just overwrite files? EDIT: I uninstalled and overwrite old files with new files and installed again. Will report back if the problem is still there.
question 2: I have defined that sd contest should use xxx madvr, and 720p yyy etc. but it is all defined in Kodi and not in madvr. Is this okay?

As long as the Kodi gui is enabled, those profile rules will override any set in madVR.
Reply
FYI, If anyone is curious if their display is capable of receiving 4:4:4 chroma subsampling from a HTPC, I found a great test image below. Most TV's are only capable of 4:2:2 from a 4:2:0 source image, but some will do 4:4:4.

HTPC Chroma Subsampling: (Source) Y'CbCr 4:2:0 -> (madVR) Y'CbCr 4:4:4 to RGB -> (TV) Y'CbCr 4:2:2 or Y'CbCr 4:4:4 or RGB to RGB

4:4:4 Display Support Test Image
Drag image into MPC-HC window; Support determined by the numbers most clearly visible: 4:2:2 or 4:4:4.
Reply
I also updated the section in the guide dealing with image doubling:

4K Scaling & Image Doubling

What if you are scaling content to 3840 x 2160 (4K)? The example above assumes all content is being scaled to 1920 x 1080 (1080p). It also assumes most resizing takes place between 720p and 1080p. Let's look at this and three other common resizes comparing the pixel density (increase in pixels) and scaling factors (increase in pixels per inch) involved:

720p -> 1080p
1280 x 720 -> 1920 x 1080
Increase in pixels: 2.25x
Scaling factor: 1.5x

SD -> 1080p
640 x 480 -> 1920 x 1080
Increase in pixels: 6.75x
Scaling factor: 2.25x

1080p -> 2160p
1920 x 1080 -> 3840 x 2160
Increase in pixels: 4x
Scaling factor: 2x

720p -> 2160p
1280 x 720 -> 3840 x 2160
Increase in pixels: 9x
Scaling factor: 3x

Resizing from 720p to 1080p represents a scaling factor 1.5x, while scaling from 1080p to 2160p doubles the resolution by 2x. In situations involving large scaling factors (2x or greater), it may be beneficial to use madVR's image doubling. Image doubling does just that – it takes the full resolution luma and chroma information and scales it by factors of two to reach the desired resolution (2x for a double and 4x for a quadruple). If larger than needed, the result is interpolated down to the target.

Doubling a 720p source to 1080p involves overscaling by 0.5x and downscaling back to the target resolution. Improvements in image quality may go unnoticed in this case. However, image doubling applied to larger resizes of 480p to 1080p or 1080p to 2160p will, in most cases, result in the highest-quality image.

super-xbr vs. NEDI vs. NNEDI3 Image Doubling

Three options are available when image doubling: super-xbr, NEDI & NNEDI3. Unlike linear scalers such as Jinc and Lanczos, image doubling algorithms rely on pattern recognition through statistical sampling. The performance of these algorithms can be slow because several calculations are made to upscale the image resulting in additional GPU processing time. Image doubling algorithms are most effective when applied to resizes at least 2x or larger. Incremental improvement may be observed in smaller upscales, but the corresponding resources consumed upscaling and downscaling may not be worth the extra processing.

super-xbr
  • Resolution doubler;
  • Relies on RGB inputs - luma and chroma are doubled together;
  • Fastest of the three. Slightly faster than Jinc;
  • Sharper than Jinc with less ringing;
  • Less aliasing on edges than NNEDI3 16 neurons;
  • Best bang for the buck.
NEDI
  • Resolution doubler;
  • Relies on RGB inputs - luma and chroma are doubled together;
  • Second fastest of the three. Slower than super-xbr;
  • Least sharp of the three. Best used with SuperRes;
  • Known to introduce artifacts when used alone.
NNEDI3
  • Resolution doubler;
  • Uses YCbCr color space - capable of doubling luma and chroma independently;
  • Slowest of the three. Slower than NEDI;
  • Similar sharpness to super-xbr. But more in-focus;
  • Best overall image characteristics - sharpness, aliasing and ringing.
Image Comparison – Clown:
Jinc
NEDI
super-xbr
NNEDI3 256 neurons

Image Comparison – Lighthouse:
Jinc
NEDI
super-xbr
NNEDI3 256 neurons

Image Comparison – Lighthouse Top:
Jinc
NEDI
super-xbr
NNEDI3 256 neurons

Note: Chroma upscaling is a form of image doubling, and all three algorithms are available for this purpose. Visual differences between algorithms will be small when upscaling the chroma layer alone.
Reply
I'm still having problems with the interaction of refresh rate switching and Reclock having trouble detecting the correct refresh rate of my TV (this never happens when using MPC-HC).

I'm using a GTX 660. The TV is capable of 23, 24, 50 and 60Hz.

Tried MadVR exclusive, tried MadVR overlay, tried Kodi Exclusive, tried Kodi windowed... one way or the other results are never perfect. Anyone with a perfectly working combination?

Sometimes I find the Reclock icon yellow, sometimes MadVR states that composition rate is 60Hz, sometimes it doesn't. All seemingly random, even on the very same file played back once and then twice, leading to different CTRL+J stats...

Often Kodi exits from playback and, although it has switched back to 60Hz, it renders the interface at 24fps.
For troubleshooting and bug reporting please make sure you read this first (usually it's enough to follow instructions in the second post).
Reply
  • 1
  • 56
  • 57
  • 58(current)
  • 59
  • 60
  • 532

Logout Mark Read Team Forum Stats Members Help
Kodi DSPlayer – DirectShow Player for Windows47
This forum uses Lukasz Tkacz MyBB addons.