If Training Mode or Similar Modes Had a Snapshot Feature

[Devon]

All right, I bought the Elgato device from Amazon and it's amazing, but there's no way to plug the HDMI to the device due that damn HDCP, so I've got to use component.

The quality it's far that the one you posted here, but in any way it's quite good. Some tests;

omg hi :D can you take a screen shot of Tira.. I just wanna see her in HD gameplay mode

 

[MetisConnetis]

I'm trying to get better quality, by now I'm afraid this it's the best quality I can get.

Here, your Tira pics ;)

 

[skytoast]

Well the images are a bit "soft" but it isn't that bad. My guess for the quality is due to the compression codec it uses which I am not familiar with. If it compresses directly to something akin to H264/AVC mpeg4 (on the fly) then you won't get the visual fidelity. Whereas the Blackmagic capture devices (or equivalent i.e. Aja, Matrox, etc) are able to capture uncompressed video with a full color gamut of 8 or 10bits. It shouldn't really matter whether your HD source is component or HDMI since I am able to reproduce the same sharp quality using component inputs on the Intensity card from an HD source (i.e. PS3).

However, I should mention when I capture video I don't necessarily use pure uncompressed HD format due to the size of the files. [One minute of pure uncompressed HD @ 720p/59.94fps is roughly equal to 6.4 GB - Yes you read that right: 6.4 Gigabytes! One hour of uncompressed HD at the same resolution and frame rate would = 375 GB!]

So instead I do use what is called an "intermediate codec" for editing video which is often used in the industry during post production work since it has the advantage of saving space (for archiving master) and unifying formats across a production workflow usually without compromising quality. An intermediate codec often times offers an easier format to work with since it involves "less" compression than something like a distribution codec for media like H264, MPEG2, divx, xvid, etc.. which is highly compressed and would require more computer resources to handle during editing (given at the equivalent bit rate, resolution, etc).

The codec I use is Cineform in my workflow on the PC (mostly with Adobe Premiere/After Effects) which is great since it cuts the file down to about 1/3 of the original and is virtually identical to uncompressed HD to the human eye thanks to its proprietary wavelet compression. It also offers other benefits like real time on the fly instant color correction without having to render anything. I can change the hue, brightness, saturation and such without processing anything which is a huge a time saver. It works in tandem with the Blackmagic cards (also supports AJA cards as well).

Here's a still comparison between Cineform and Uncompressed HD. Can you tell if there is any difference?

Cineform HD 720p 59.94fps 8 bit 4: 2: 2 YUV (medium settings)

Uncompressed HD 720p 59.94fps 8 bit 4: 2: 2 YUV (raw)

Both of these are a single frame in the video. Even using the Cineform codec at medium settings still yields a beautiful image almost indistinguishable from the original. I wish I could show you guys the real video footage. It basically looks like a pure stream from the console. Every frame in the capture video is this sharp like the still image above. Putting it up on sites like Youtube cant do it justice since it requires encoding it again to a different format (distribution codec mentioned above like H264) which again gets re-encoded even further from their end.

Here's some more images using Cineform:

 

[MetisConnetis]

It was difficult to find accurate information about the capture devices, besides all commercial stuff.

My first choice was the Intensity Shuttle, after you gave me the info about Blackmedia, but works only with USB 3.0, and my laptop has 2.0. So, my last option was ElGato, and yes, I'm afraid it has chip-based video compression, and that's the difference between your cristal-clear pics and my blurry pics.

The codec you used to compress the video it's nearly perfect, I wish I could choose another video compression to avoid that "blurry" effect, but I guess that's the best I can get without a capture card or a modern laptop...

Actually I'm not using the wire that Elgato gave me for the PS3, because my TV it's old and has just a DVI connector (so I wouldn't have any sound if connected), so I'm using the PS2 component cable. Somehow, this makes the screen even more blurry (due the Elgato switch, if plugged directly the PS3 with TV it's a very clear screen), so maybe, when I start using the right wire (I think that night I'll have the Digital-Analog Audio converter to split video and sound from PS3), my pics we'll be a little more clear, but, I know, nothing like the ones you're posting here :(

Thanks again for all the info ;)

 

[Norik]

That's awesome! How does the video look? Any chance of a little upload to see?

:3

 

[skytoast]

Hmmm.. I guess I can do that.

OK here. https://www.dropbox.com/s/n61i03prqgmhe22/Cineform sample test - SCV Viola.rar

In order to view it, you'll need to install the Cineform decoder codec which is free. You can get that on their official page here (available for both PC and Macs):

http://cineform.com/gopro-cineform-decoder

It's only a decoder meaning allows you to playback Cineform encoded files. If you want to encode or make your own Cineform files, you'll need to purchase their license which starts at $299 (Cineform Studio Premium. The other cheaper Neoscene at $129 only encodes HDV/DSLR, with no capture support). It's gotten considerably cheaper and more affordable since I first started editing in HD back in 2006. The license then use to cost $1000+. You'll notice they are now a division of GoPro (they make those awesome little wearable sports/action HD cameras) as they got acquired by that company in 2011. The original team still works with it and have improved it greatly over the years, adding real time color correction metadata and 3D support.

Anyways, the video sample above is a 00:03:27 sec clip of Viola encoded in Cineform @ 1280x720 [720p native resolution] 8 bit YUV 4: 2: 2. Its about 59 MB in size. The same 3 1/2 sec clip in uncompressed HD would = about 423 MB. (Yes for only 3.5 seconds). That's a dramatic reduction in size but the quality is virtually indistinguishable to the human eye.

After viewing it, compare it with the still frame above. =)

 

[Norik]

That. Is. AWESOMEEE!!!

 

[MetisConnetis]

That's freaky awesome!!

With the right wire pluged all I can get it's a slightly cleaner screenshot, nothing like that.

Taking a similar pic, this is the difference;

BEFORE;

AFTER;

As I said, it seems a little cleaner, but I'm afraid this it's the best I can get...

By the way, the codec you posted, it's the one used in the GoPro cameras?

 

[skytoast]

GoPro uses H.264 mpeg4 codec. Cineform was bought out by them in 2011. Depending on which model, resolution can go all the way to 4K resolution such as with the new Hero 3. I'm not really sure what bit rate is availble in recording on GoPro cameras. Cineform as intermediate codec is currently still too big for something portable like that even though the files are much much smaller than uncompressed formats, hence being an "intermediate codec". There are cameras like Silicon Imaging SI-2K which is a professional digital film camera that can utilize Cineform Raw. Some movies were shot this way such as "Slumdog Millionaire". Of course, the other popular digital film camera is made by RED with its own intermediate REDcode. If you got Final Cut Pro on a Mac, there's an excellent intermediate codec with Apple ProRes 422 or 4444. They all handle everything from SD NTSC/PAL to HD to 4K resolutions and may also handle 3D as well.

With GoPro, I believe you get special access and utilities with the free version of Cineform Studio software with each camera that enables you to convert the files to Cineform (should you wish). The advantage of course is the extra metadata handling such as color correction I mentioned earlier and the ease of use. DSLR from Canon and Nikon can also be converted to Cineform for ease of handling as well. The difference with getting a full license is software integration for post production workflows (codec handling in applications like Adobe, Avid, Sony Vegas, etc) and of course hardware interfacing like capturing and conversion.

With the Blackmagic capture devices, you dont really need to go out an get an intermediate codec like Cineform if you dont want or have the need to. They include the MJPEG (Motion JPEG) codec and the software has full integration for NLE (Non Linear Editing) packages like Adobe Premiere with project presets and capture ability right within the NLE. For most purposes this would work very well as an intermediate codec as it also cuts the file size down. It lacks all the bells and whistle such as on the fly color correction and 3D metadata handling but should suffice for most uses. Best part, its free included in the package. =)

Blackmagic's implementation of MJPEG looks just as good as uncompressed video. There's slight artifiacting around edges and some less color fidelity but for videogame sources, you wont notice that much of a difference. Artifacting use to look much worse on older revisions but BM has made improvements along with better handling from modern multi-threaded computers.

Here are some comparison shots against Cineform:

MJPEG HD 720p 59.94fps 8 bit 4: 2: 2 YUV

Cineform HD 720p 59.94fps 8 bit 4: 2: 2 YUV

MJPEG HD 720p 59.94fps 8 bit 4: 2: 2 YUV

Cineform HD 720p 59.94fps 8 bit 4: 2: 2 YUV

Looks good, doesn't it? Even with the flashing text in the corner, doesn't look bad with no real visible aliasing issues. There's some slight color differences but overall looks great.

I haven't looked much into the Elgato device but the clarity of the video is probably due to the limitation of the hardware and how it handles its analog to digital conversion.

------

For kicks, added some extra shots to match your screenshots using BM+Cineform.

 

[Norik]

Damn, now I want one myself...

By the way, my computer ran the video without having to install anything (it's a mac, maybe that's why).

 

[skytoast]

hmm.. really? Interesting. I wonder if the decoder is somehow being distributed with the OSX now or some other package. Either that or you installed it through something else that probably had the decoder in it. I tried opening a CF file on a spare laptop that didn't have the CF codec installed using the latest version of Quicktime but all I got was audio and it was still asking for a compatible video codec.

 

[MetisConnetis]

I see this;

And this;

And I want to cry! :(

Thanks skytoast for all the info ;)

 

[skytoast]

I wouldn't worry too much about it. I read a small review on the Elgato from another user in this forum in the Technology section http://calibur.8wayrun.com/threads/elgato-game-capture-hd-review.15151/

And for what it is, it isn't that bad for most people's uses especially if you have limited hardware and all you want to do is share videos on something like Youtube. In the article linked above, there's an example of captured footage using the Elgato on Youtube. Watching the video at low standard setting of 360p-480p, the video looks awful. However setting the video on Youtube's 720p HD setting actually looks pretty good for streaming. (Of course this all depends on what bit rate ratio and format you send to Youtube, only to have them re-encode it again on their servers.) This begs me to ask if you you can up the bitrate when you capture video on the Elgato. Is there something like a quality setting you can adjust?

Even my videos can look awful on Youtube with low settings. The HD setting of 720p-1080p is acceptable but as apparent no where near the gorgeous picture from the original.

Here's the same clip of Viola I uploaded earlier only this time converted from its original Cineform format to something Youtube could recognize. H264

H264 720p High Profile Level 5.1 Bit rate 20Mbps CBR (Constant Bit Rate)

At the low settings of 360p / 480p, its obvious it looks like crap. Adjusting the setting to 720p playback is somewhat decent. Unfortunately this is as good as it gets (for now), and when compare it to the original video you make at home/studio you do want to cry... lol.

Perhaps someday in the future when technology catches up and network bandwidth is ridiculous high (and cheap and affordable), we could all be streaming uncompressed or near uncompressed formats. But by then I'd imagine we be dealing with other challenges like holography and spatial immersion...

But for now, to properly setup a rig for true HD capture like the above, you'll still need proper interfaces, a fast multicore (4 or more) processor environment, and lots of speedy hard drives connected in parallel in RAID (to keep up with the high bandwidth requirements for HD or better to write to disk) and most likely some expensive software and plugins to go along with that. I can't really see most people doing this unless you do this for a living, an enthusiast or really serious about video editing and such. So the device like the Elgato should do.

 

[MetisConnetis]

That's the point, you're right, the quality I want it's something like professional edition but afforadable, and it's a nonsense. Devices like HDV, Roxio, Avermedia o Elgato offer a very good quality without previous knowledge of video edition. As always it has been, the professional tools worth it... as you said, if you love/work with it.

The software from Elgato has a quality slider. The pics that I took was configured to "Max".

 

[SophiRprsnt95]

Here's some snapshots from SC II! :o)

It's NOT a glitch. I was playing as Taki and did some silly things like this.

 

[Schneider-X]

That's the point, you're right, the quality I want it's something like professional edition but afforadable, and it's a nonsense. Devices like HDV, Roxio, Avermedia o Elgato offer a very good quality without previous knowledge of video edition. As always it has been, the professional tools worth it... as you said, if you love/work with it.

The software from Elgato has a quality slider. The pics that I took was configured to "Max".

Have you tried converting the video in the " EGC_library?" The video there seems to be based off of the flashback recorded videos that the software captures. But this video is in the MPEG-2 TS format, I'm not sure if that's a lossless format or not, but the file size is significantly higher compared to the MP4 videos created. Maybe you can apply Cineform to this and have a clearer picture.

The downside to this method is that you will have much more video to convert and process instead of what you just want recorded.

 

[skytoast]

Applying Cineform won't "improve" quality from the source but rather give you a "lossless" intermediate video equivalent that's easier on CPU resources when decoding. I'm not exactly sure what the specs are on the Elgato device since I cant find any published literature on the specs it uses for capturing video. Most likely however I suspect that since its a consumer level device, it captures video with a color space of 4 : 1 : 1, 4 : 2: 0 or maybe 4: 2: 1 which would explain the lack of color fidelity when compare to higher end capture solutions such as from Blackmagic-Design, Aja, Matrox, etc. With the higher end cards, video is captured in 4: 2: 2 color space.

I'll explain what these ratios mean in a second but first I need to briefly explain how the human eye percieves color. The human eye uses rods and cones to percieve light. Rods are sensitive to levels of light intensity (luminance). Cones are sensitive to spectrum of color. There are more rods in the human eye than cones therefore we are more sensitive to changes in illuminence, brightness/darkness, black/white, shades of grey than changes in color. Because of this, we have come up with ways to remove redundancy in imaging technology by sub-sampling or removing certain aspects of color to make it more efficient. This makes it easier when we want to produce, broadcast, transmit videos and still images in things like cameras, recording devices, encoders, TV/display devices, etc.

Typically, color is displayed using three primary colors: Red, Green and Blue which is commonly referred to as RGB color space. Mixing any of these colors to a greater/lesser degree could yield any color in the visible spectrum. This is similar to mixing paint on palette (except in reverse -mixing colors with light is additive -mixing with paint is subtractive). Old CRT tube TV worked this way using three electron beam emitters that would hit a phosphor screen that was laid out in a grid with each area having pigments that contain Red, Green and Blue "zones". Depending on the intensity and direction of the beams, the phosphor pigments corresponding to each color would "react" in such a way and produce the desired color of light. From a distance, we see a combination of these "flashes" over a given period of time (fractions of second) as images that we see on a TV screen. This is still similar in today's HD TVs using LED/LCDs except instead of using scanning electron beams that hit a phosphor screen, we use an array of tiny light emitting transitors made of liquid crystals with each of the three primary colors in our glass displays.

When we zoom in on an image, whether it be a still image or moving video, we find that its made of fine pixels grouped together in many shades of color. Usually what we percieve on a screen as only one color like someone wearing a red sweater is actually made of many different colors and shades when zoomed in. There are several ways we can divide and group color and translate that to the appropriate media by association.

RGB color representation is most commonly used in imaging such as digital photos and print media. Since we usually can scrutinize more details in still images, this straightforward method of utilizing three colors offers the best method giving the best quality of picture. Thus you see this color scheme commonly used in imaging applications like Photoshop, where often detail is refined at pixel level. However since in video, we are dealing with many images in motion over a very brief span of time (usually 1 sec -known as "frame rate"), it is not necessary to have this level of detail since our eyes are not very sensitive to minute changes in color fields. Since our eyes are more sensitve to levels in luminosity, video standards typically use the color field of Y,Cr,Cb or commonly refer to as YUV. This standard uses one component Y that represents the level of brightness/darkness and two component of color (chroma) along the horizontal and vertical resolutions between RED and BLUE. By controlling the level of intensity in luminosity along with shift between two chroma spectrums of red and blue, its possible to make an approximation in which our eyes will percieve as any color within the light spectrum.

Imaging devices like cameras and capture cards that use YUV color schemes, will sample a small reference area (typically 4 pixels wide) for each component. Thus video color spaces are represented using 3 ratio numbers such as 4: 2: 2, 4: 1: 1, etc. The first ratio Y is left alone since our eyes readily percieve changes in brightness and darkness. Colors however can be omitted in the remaining two samples in UV (Cr, Cb) space, the remaining colors (if any) averaged together and still offer the illusion of the percieved color representation by way of approximation at the cost of color fidelity.

A perfect image without any omission in colors would be represented using 4: 4: 4 color space. In this scenario, we have 4 full pixel sample for luminosity, 4 pixel samples for 1st Chroma value in Cr space and 4 full pixel samples in 2nd Cb space. This is rarely used except in very high end digital camera equipment such as cinematic RED digital camera used for feature films shot in large 2K-4K resolutions (bigger than 1080p HD). 4: 2: 2 color space today is the "gold standard" for most high end videos. A very clean true image can still be retained by only omitting 2 colors from two sub components. In a 4: 1: 1 color space, we start to lose some fidelity because more color components are removed from the chroma space leaving only a single color for representation.

In the old days of video editing, I used an external capture card by Canopus over firewire for DV formats. This device captured video using a color space of 4: 1: 1. This wasn't much of problem back then because TV screens were of lower resolution having smaller dimensions than today's 1920 x 1080 pixel wide screens (1080p). Since TVs were mostly CRTs, video signals were interlaced and "scanned" across the screen. This is where the image was made by "stitching" two overlapping beam projections "scanned" horizontally in different directions (inter-laced), pulsing or flashing at a certain rate depending on your region of viewing (NTSC/PAL) across the screen. This action along with smaller screen resolutions masked much of lack of detail. With the advent of higher resolutions in HD displays/devices and non-interlaced progressive images being the norm, there is more incentive to capture more color fidelity in video images.

So I'm afraid you can't really do much to improve visual quality with the videos you capture using the Elgato device. It's really dependent on the hardware components inside. This isn't to say the device isn't good and as I've already mentioned, should be enough for most users who want a simple capture solution for home viewing and over the internet streaming.

 

[SophiRprsnt95]

I'd really love to get all this high-tech stuff.

Anyway, here's a new addition to this humorous gallery. :3

Now for a closeup! :3

Alpha Patroklos is NOT amused. o_o

These obviously aren't in the best quality, but I try. For this thread, I prefer some more humorous snapshots.

 

[Devon]