|
|
![]() |
||||||||||||||||||||
|
Best Blu-ray Movie Deals
|
Best Blu-ray Movie Deals, See All the Deals » |
Top deals |
New deals
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
![]() $24.96 17 hrs ago
| ![]() $44.99 | ![]() $31.13 | ![]() $24.96 | ![]() $19.99 10 hrs ago
| ![]() $70.00 | ![]() $29.95 | ![]() $20.07 7 hrs ago
| ![]() $54.49 | ![]() $27.13 1 day ago
| ![]() $29.99 1 day ago
| ![]() $34.99 |
![]() |
#1 |
Blu-ray Ninja
|
![]()
With the recent news and statements made by Toshiba, certain things have upset and enraged everyone at the Blu-ray community. Mainly I am talking about their claim that they can make a DVD upconverter that can give you an image equal to an "HD DVD" -or rather Blu-ray or HD- image. Well I decided to investigate this claim more fully than just not accepting the logic that an SD image just doesn't have the information to make an HD image.
So I decided to do a little research. Unfortunately I am not able to copy or find the exact results of a DVD upscaler of the quality they talk about, since the technology simply is not out there yet. Sure there are upconverters, but none close to true HD quality. So I used the next best thing to experiment with upconverting an image by myself. In the Digital Photography and Image Editing world, there are certain standards everyone uses. For high end professionals like myself (we are talking about "National Geographic Society" level here, I do work there), Photoshop is a must for image editing. However there IS a standard superior to Photoshop for Image Enlarging. This program is called "Genuine Fractals". Now this is mostly used to enlarge an image using a logarithm that is similar to a fractal to interpret the missing data. This is far more accurate than the standard "bicubic" resampling which merely copies pixels and blurs or sharpens edges. As far as I know, Bicubic is the only way upconverting DVD players work, but if someone has more information on this please share your knowledge so I can do a more extensive study. Genuine fractals is used mainly for prints, but can also be used for converting SD video to HD (they include a tutorial with the software), but it takes several hours for even a short video. Either way, the important thing to note here is that Genuine Fractals takes approximately 5 seconds on a high end computer to process each image, while bicubic takes closer to 1.5 seconds. Either time is too long for a DVD/video which has to process anywhere from 24 pictures each second to 60! Still, assuming that somehow Toshiba was capable of harnessing the processing power of a high end computer, perfecting the logarithms of fractals, and cutting down the required time to convert such images, the following would be the results and comparison to what you would see. The first image is the actual data on an anamorphic DVD. On the internet it looks too tall and misshapen, which is due to the anamorphic nature of the pixels (this will show up ini Photoshop). The second image is what you will see on your SD TV from a DVD. The third image is the actual source, true HD at 1920x1080. The fourth image is the upconverted result using genuine fractals (industry standard) to make an "HD" image. UPDATE: The fifth image is what a bicubic up-sampled image will look like. Again I am not sure the technique used for upconverting DVD players, but this is the best I can reproduce using high end digital imaging. ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Pay special close attention to the details in the edges, the fine grain and the wood textures. What do you all think? Can Toshiba make a true HD image from a DVD? I didn't think so. NOTE: To enlarge an SD image to a FULL HD 1080 image, you need to enlarge the picture ~267%. This does not include bandwidth, sound, motion, interlacing/progressive. WIth even a single image it is impossible to get close to true HD from SD no matter how advanced the technology you use. Last edited by CptGreedle; 03-05-2008 at 05:57 PM. |
![]() |
![]() |
#5 | |
Active Member
Dec 2007
Staten Island NY
|
![]() Quote:
![]() |
|
![]() |
![]() |
#6 |
Active Member
Oct 2007
|
![]()
Yes, totally agree... what about motion? The real video... Yes for us, here at Blu-ray.com, we like the perfect HD images we get with BDs. But what about J6P?? Or those little families with a small collection of 50 or 100 DVDs?? Would they care about PQ?? If Thos will try to sell them the "perfect" HD image for half price of a Blu player.. What is the idea of Thos? To keep messing with Blu-Ray as much as they can??
Best, DJ Headd |
![]() |
![]() |
#7 |
Senior Member
|
![]()
anyone with half a brain cell would know that Toshiba's upconversion claims were pure BS, but it's good to see examples anyhow.
why not include a 5th image showing an upconverted image using a bi-cubic resampling since that's essentially the best that can can be reasonably expected of upconverting players? |
![]() |
![]() |
#8 |
Blu-ray Ninja
|
![]()
Motion is indeed a different aspect that I can not reproduce right now. But I will work on it. I can take a short sequence and upconvert it to show the differences, but it will take some time.
Also, another problem is dealing with bandwidth. DVDs do not have the bandwidth of a BD and the upconverting players will not be able to make that up so you are limited in bandwidth as well but I am not sure by how much. I understand that you think J6P won't see the difference. But just like edge enhancement, once you see it, you hate it. Sure they can watch their DVDs, but when they get a big screen and see the quality difference, I don't think anyone will believe that they look the same. |
![]() |
![]() |
#9 | |
Blu-ray Ninja
|
![]() Quote:
|
|
![]() |
![]() |
#10 |
Expert Member
|
![]()
As I said in another thread, we need no more than simple logic to see that it cannot be done. If you can go from 720x480 to 1920x1080 and produce an identical image, then you can do the same again and go to 4k, and then onto a gigapixel image etc. Indeed why start at 720x480, start at a single pixel and upscale it to 6 pixels then 36 pixels etc. If you can always upscale to 6x the pixels and get an identical result to an image natively of that resolution, you can keep going to 1920x1080 from a single pixel. Clearly this cannot be done, anyone should be able to see that.
|
![]() |
![]() |
#11 | |
Blu-ray Ninja
|
![]() Quote:
Sure the logic is there, but without the hard evidence to show people what's the point of the argument? Enough bad logic and denial went on during the format war, lets not repeat that again... |
|
![]() |
![]() |
#13 |
Senior Member
Sep 2007
|
![]()
It's easy to explain using Shannon's law why entropy in a closed system cannot increase and hence explain why you cannot create missing data for frames outside of simple geometric objects in any size constrained heuristics with a bound on information it has.
But people have been selling snake oil on upscalers, deinterlacers for so long now that many just take it for granted that there are tricks available that can recreate missing details. The problem is simple. If you have a photograph, and you cut it into tiny horizontal and vertical strips of equal width and height, say you get 1920x1080 tiny squares (only one color in each square) that make up each picture. Now if you can only pick 720x480 squares placed at relatively equidistant to rebuild the whole frame. You throw away the other squares. Once you only have that 720 x 480 squares (each of which is a single color), you cannot recreate the missing squares that was thrown away. People fundamentally understand this, yet they are also fooled by claims of information creation through super scaling or through motion interpolation when they really should have known better. |
![]() |
![]() |
#14 | |
Blu-ray Samurai
May 2007
Indianapolis
|
![]() Quote:
HOWEVER, this thread brings up a question. With the Genuine Fractals technique, wouldn't it be possible to convert an older movie to HD? |
|
![]() |
![]() |
#15 |
Blu-ray Ninja
|
![]()
Couldn't have said it better myself. However something you didn't talk about is that you can program an algorithm to guess what the lose information would have been based on what is there. THAT is the best and really the only way to enlarge an image if you have no alternative.
But like you said, it is impossible to bring back the lost image. A guess is usually wrong, and using the best guessing algorithms available, I have SHOWN how far off it still is. |
![]() |
![]() |
#16 | |
Senior Member
Sep 2007
|
![]() Quote:
Even Toshiba is not claiming that their DV camcorders are good enough and noone should buy their highdef HDD camcorders (which noone is buying anyway, but that's besides the point). |
|
![]() |
![]() |
#17 | |
Blu-ray Ninja
|
![]() Quote:
You can take an SD video and make it look better than any other upconverting method because it spend the time to figure it out. But this would be the only viable option if there is no HD source. Since there is no HD image, there is no way to gain the missing information so it would not be "true HD" but it would be closer than Toshiba's upconverters that have to "guess" (or rather "resample") an image in real time. It will never look as good as True HD, and it will take hours or days to up-sample. |
|
![]() |
![]() |
#19 | |
Blu-ray Samurai
May 2007
Indianapolis
|
![]() Quote:
|
|
![]() |
![]() |
#20 | |
Blu-ray Ninja
|
![]() Quote:
|
|
![]() |
|
|
![]() |
![]() |
![]() |
||||
thread | Forum | Thread Starter | Replies | Last Post |
HD Test Video Test Patterns on a BD | Blu-ray Movies - North America | rlmbluray | 1 | 05-04-2009 04:35 PM |
Identifying Imaging, Presence, Depth, Soundstage and the like! | Audio Theory and Discussion | Johnny Vinyl | 9 | 02-25-2009 06:40 PM |
"Batman" film technician dies in accident | Blu-ray Movies - North America | Petey Pablo | 11 | 09-26-2007 04:12 AM |
|
|