GB2373947A - Film scanning and image compression at high resolution. - Google Patents

Abstract

A system for producing digital image files of comparatively low data size for transmission of storage comprises inputting images of high resolution but with low contrast ranges between adjacent pixels and compressing the data files to reduce the size. The system is particularly applicable to scanning film 1 at a resolution approximate or exceeding the resolution of the film, e.g. 4K, thereby producing high resolution image files for compression. By scanning film at a higher sampling resolution (as compared with standard 2K sampling), the differences between adjacent image pixels, such as edge transitions, are reduced thereby reducing compression artefacts.

Description

FILM SCANNING AND IMAGE COMPRESSION FIELD OF THE INVENTION The present invention relates to the scanning of film frames to acquire a digital representation of the film, and also to the compression and decompression of digital images.

BACKGROUND OF THE INVENTION Film scanners are well known and many such scanners exist for scanning 35 mm movie film and other film stock. One such known film scanner is the C-Reality *"telecine manufactured by Cintel International Limited. As is known to the skilled person, a telecine scans film and typically produces a television picture output. The current standard format of television output is known as standard definition (SD) and comprises 720 samples per line of

image and 576 lines. More recently, 35mm film is scanned at 1K or 2K resolution 1024 samples per line and. 778 lines or 2048 samples per line and 1556 lines. The subsequent data file, especially when scanning 2K is very large and requires considerable storage space and fast network ports.

At 1K and 2K scanning resolutions of 35mm film JPEG compression has been experimented with. In the image data world JPEG is used in preference to the MPEG compression algorithms, used with video images, as it is frame independent. MPEG compression relies on knowledge of adjacent frames.

JPEG compression has been found to be less satisfactory the sharper the image. Specifically where there are large signal transitions between adjacent pixels, there are

significant blocking artefacts. An example of this is in an area of high resolution where there is a sharp transition between black and white.

SUMMARY OF THE INVENTION We have appreciated the problem of storing, transmitting and receiving large amount of image data. We have also appreciated that compression such as JPEG can produce unwanted artefacts, but that with 4K scanning, compression and subsequent decompression of the image data files can actually result in an increased image quality compared to a lower uncompressed sampling scan at 2K.

In a broad aspect, the invention provides methods and apparatus for producing high resolution image data, compressing the high resolution image data and subsequently decompressing the image data and producing the image data in a chosen format. In particular, the invention provides for the scanning of film at a resolution which approximates or exceeds the resolution of the film, or the film and any intervening optics.

Subsequently, the image data so produced is compressed; for storage or transmission before being decompressed for use. The invention produces a better final image quality than scanning at a lower resolution.

The invention is defined in the independent claims to which reference is now directed. The preferred embodiment comprises a system for producing image data files from film. A film scanner scans the film at a scanning resolution approximating or exceeding the resolution of the film and produces image data files of scanned film.

Subsequently, a compressor compresses the data files. The compressed data can then be transmitted or stored as desired.

A decompressor then expands or decompresses the compressed data files ; and means is provided for arranging the decompressed data into desired alternative data or video formats. This arrangement provides for transmission and storage of data with a reduced data size, but with better quality than simply storing for transmitting a lower resolution, as will now be seen.

BRIEF DESCRIPTION OF THE FIGURES An embodiment of the invention will now be described, by way of example only, and with reference to the accompanying drawings in which: Figure 1: is a schematic representation of a system embodying the invention; Figure 2: is a block diagram of the main components of a film scanner; Figure 3: shows the spot size ratio of a flying spot film scanner; Figure 4: shows a schematic representation of a film scanner system embodying the invention in greater detail; Figure 5: shows how the invention may be used to transmit image data; Figure 6: shows how the invention may be used in the whole film transfer and post production process; and Figure 7: is a schematic diagram of a second embodiment comprising an upconversion and compression system embodying the invention.

DESCRIPTION OF PREFERRED EMBODIMENTS The invention in its broad aspect may be embodied in a number of ways. The key features of starting with high resolution image data, compressing that data anddecompressing to produce a desired image format provide the advantage of a better resultant image, but with reduced transmission and storage requirement irrespective of the image data source. However, the advantages are particularly beneficial for image data acquired by scanning film using a film scanner.

A film scanner, as is well known to those skilled in the art, illuminates film (typically 35 mm film) and captures image data using detectors to produce electrical signals representative of the image. A film scanner which converts directly to a television signal is known as a telecine, and such devices fall within the term film scanner. A more general film scanner simply converts the film image to digital data. One such known film-scanner is the C-Reality scanner of Cintel International Limited.

Referring to Figure 1, a film scanner 1 scans 35mm film, such as OCN (original camera negative) at 4K (4096 samples per line) resolution and outputs the digital image files as a series of 4K data files 2 as one file per film frame. These data files are compressed at 3, preferably using the standard JPEG compression algorithm. This compressed sequence of images may be stored at 8, for example for transmission to a remote site. At 4, the compressed image data is decompressed whereupon the 4K image data file can be recovered at 5 or a lower resolution 2K, or other resolution file can be recovered (at 6). Alternatively, and advantageously, a high or standard definition video signal can be obtained from the 4K data at 7.

The figure also shows a path 9 in which an uncompressed video signal 7a or image data file 6a is obtained directly from the film. We have observed that the quality of a 2K data file, for example, or standard definition video image taken from the 4K image data file obtained after 4K scanning, compression and decompression is actually better than that obtained by taking a lower resolution, say 2K file or standard definition signal straight from the film scanner.

The benefit of the system described may be appreciated from the following discussion. 35mm OCN film and/or a combination of OCN film and the camera taking lens has a resolution which deteriorates at around 4K. As 4K scanning is little used in the industry, previous JPEG compression tests have been conducted up to 2K. These have produced considerable amounts of compression artefacts which have made them unacceptable. It had been assumed that still higher scanning sampling would simply produce worse artefacts.

Compression artefacts are produced at low resolution sampling as the fixed sampling can gather in one fixed sample a dark part of the image and, in the adjacent fixed sample, a light part. There is thus a very high and rapid change in frequency (or sample level) from one fixed sample to the next, above the ability of the compression algorithm. The result is an artefact at any point in the image where there is a sharp transition edge.

At 4K, the resolution of the film is deteriorating, if one takes the effect of a camera lens, which images the film into account. Adjacent pixels, if sampled at this higher sampling resolution, will have little difference between them. That is, the edge transition at the higher sampling density will now be smoothed over a number of pixels. As a result, the JPEG compression process will not produce as

many artefacts at the lower sampling density for a similar image area.

When scanning 35mm film at 4K, the compression rate may be typically 60: 1 which generates file sizes of about 1 Mbyte per frame. Such files have been found to have a quality which far exceeds the quality of uncompressed 2K files at a file size of about 10 Mbyte per frame.

By scanning at 4K and compressing aggressively, such as up to 60: 1, full resolution files may be transmitted easily over very cheap, low grade networks, stored and subsequently decompressed allowing full manipulation and/or offline edit versions. This enables a remote user to start work on a film even while it is still being scanned.

The example described uses 4K scanning with 35mm OCN film.

Different scan and sampling rates will be appropriate with different film stocks and formats. However, the improved performance is achieved by scanning the film at a resolution which approximates or exceeds the resolution of the film image, preferably when any imaging lenses are taken into account.

The main functional components of image capture of a cathode ray tube (CRT) film scanner and CCD film scanner are shown in Figure 2. These functional diagrams demonstrate the resolution of an image that can be achieved with film. In the first example, a light source 10 illuminates film 12, and the film image is presented to CCD detectors 18 via a colour beam splitter 16..'In such a CCD film scanner, the resolution of the captured image is determined by the resolution of the CCDs 18. The second, preferred, example is a CRT film scanner using a cathode ray tube (CRT) 20 to provide a flying spot of light onto film 24 via imaging optics 22. The image at each point on

the film is then captured on detectors 28, again via a beam splitter 26 to provide separation of light into separate red, green and blue components. The resolution of the signal is not dependent upon the detector resolution, because the detectors simply need to capture all light transmitted by a particular point on film. It is the spot size which determines resolution as shown in Figure 3.

The CRT 20 has a dimension 30 across its face of around 100mm. This compares to the film width 31 of 24mm. Now the size of spot of light on the CRT raster face is around

30um giving a resultant spot size on the film of 30um x (24/100) = 7um. Thus across a 24mm film frame, there are 24mm-7um = 3, 400 contiguous spots which is approaching the 4K scanning actually achieved. In fact, 35mm camera film is capable of providing a resolution of 6K, but the camera lens through which the film is exposed reduces this actual resolution to 4K or possibly 3K. Furthermore, subsequent transfer of film negative to positive and production of further copies for distribution reduces this resolution to 2K. It is for these reasons that it is recommended to scan directly from OCN film at 4K sampling rates to achieve the best image quality.

As the resolution of a scanned image increases, however, so does the size of each file as shown by the table below setting out the file size in Mbyte.

Mbyte for each: Frame Second Minute SD 1.6 38.4 2.3Gb 1K 3.2 76.8 4.6Gb HD 8.2 197 11.8Gb 2K 12.5 300 18. OGb 4K 50 1. 2Gb 72. OGb

As can be seen, the amount of data to be stored and transmitted rapidly becomes large. The size of files is addressed by the system embodying the invention as shown in Figure 4. The film scanner system 41 comprises a telecine or film scanner 32, an example being the commercially available C-Reality scanner which colour corrects and produces 4K image files from film.

Subsequently, postware 34 was used to store the files in 10 bit log format prior to linear conversion by a linear converter at 36. Further colour correction was then performed at 38. The scanner system then has a compression function 40 which is preferably a standard processor running the known JPEG compression algorithm.

Two outputs are provided: an uncompressed output 45 producing uncompressed 4K data 42, and a compressed output 43.

The maximum data rate output in frames per second for various different file sizes is shown in the table below.

Frames per second 100BaseT Hippi Fiber LVDS HSDL GSN SD 6.3 43.8 62.5 70 188 500 1K 3.2 21.9 31.3 35 94 200 HD 1.2 8.5 12.2 13.6 36 98 2K 0.8 5.6 8.0 9.0 24 64 4K 0.2 1.4 2.0 2.3 6.0 16 As can be seen, even the fastest data port only reaches 16 frames per second-below the film taking rate of 24 frames per second. The compression function 40 using JPEG can vary the amount of compression applied. The tables below show the frame rates and storage space required for different compression factors.

Compressed 4K Frame Rates (16 bit) 4K 4K 5 : 1 4K 10 : 1 4K 26 : 1 4K 52 : 1 100BaseT 0.1 0.5 1.0 2. 6 5. 2 Hippi 1 5 10 26 52 Fiber 1.3 6.5 13 34 68 GSN 10 50 100 260 520 Storage for 4K compressed files (Based on 75 Mb uncompressed @ 16 Bit) 4K 4K 5 : 1 4K 10: 1 4K 26: 1 4K 52: 1 30 sec 54Gb 11Gb 5.4Gb 2.4Gb 1.2Gb 30 min 3.2Tb 640Gb 320Gb 123Gb 61.5Gb 60 min 6.4Tb 1.2Tb 640Gb 246Gb 123Gb 120 min 12.8Tb 2.6Tb 1.2Tb 492Gb 246Gb The savings in storage space and data transmission are clearly significant.

A system using the invention in the distribution of image data is shown in Figure 5, and comprises a film scanner 32, postware 34 and a compression function 40 as before.

Within a computer network, the compressed data may be transmitted compressed using the compressed output 43, or could be stored in a relatively small store 52 which can serve as a master archive. Decompression functions 44 are provided as are scalors 56 which provide the data to a large server store 70. Edit functions 68 using the decompressed data in the large store 70 are provided as is further compression 58, storage 60 and decompression 62 to provide a distribution route for display. The compressed 4K data is used as a real archive so that the film only needs to be transferred once. The storage requirements and data rates are such that even a simple computer network has the capacity to store and transmit the data.

The whole film transfer and post production process using the invention is shown in Figure 6. As before, a film scanner produces 4K data files which are compressed by a compression function 40. A store 52 serves as a master archive for the film image at 4K resolution (compressed).

An editor/manipulator 74 (which are commercially available) is arranged to edit the images before providing the result to either film or television distribution. The film distribution uses a decompression function 44 as before, and then provides this either direct to a digital projector 96, or to a film recorder 90 to produce film prints for projection.

The alternative route is to use a destination formatter 78 and scalor 80 to provide a chosen television format such as digital projection 82, HDTV 84 or SDTV 86. In either case, the end result is a better quality image than if the relevant signal format had been taken direct from the film scanner 32. This is because, as previously described, the compression technique produces fewer artefacts the higher the resolution of the source from film, because the source data will have fewer sharp transitions.

The same principles apply to a second embodiment of the invention shown in Figure 7. In this embodiment, the image source 106 is preferably, but need not be, an image taken from a film scanner. The image could be a standard definition television signal (at 625 lines). In a first version, a compression system 100 comprises an upconverter 108 and compression function 110 and a separate decompression system 102 comprises a decompression function 112 and downconverter 114. In between at 120 storage, transmission or other functions may be provided as previously described. In the alternative, the compression and decompression systems could be combined to form a single storage unit 104 with storage provided at 122. Now, the synergy of upconversion and compression

provides exactly the same benefits as previously described. The resultant data files are smaller to store and transmit allowing archiving and manipulation of images with lower hardware overheads, but without significantly impacting image quality. This is because the upconversion process produces data files which have fewer sudden transitions i. e. the contrast range between two pixels of the original image is split between more than two pixels of the upconverted image, with the pixels introduced by upconversion having contrast values between those of the original pixels. When compression is applied, the reduced contrast range between adjacent pixels reduces the number of artefacts introduced by the compression and decompression. Thus, contrary to what one might expect, it is better to initially increase the amount of data thereby reducing contrast between adjacent pixels before compressing.

In summary, the embodiments provide a system in which image data is produced at a higher resolution but low contrast range between pixels prior to compression to improve the image resulting after decompression.

Claims (23)

1. CLAIMS 1. A system for producing digital image files of comparatively low data size for transmission or storage, comprising: an image input device arranged to produce image data files of high resolution but with low contrast ranges between adjacent pixels; and a compression function configured to compress the data files to reduce the size of the data files.
2. 2. A system according to claim 1, wherein the input device is a film scanner for scanning film at a scanning resolution approximating or exceeding the resolution of the film.
3. 3. A system according to claim 2, wherein the scan resolution is substantially 4K.
4. 4. A system according to claim 2 or 3, wherein the scan resolution is 4096 pixels per line.
5. 5. A system according to claim 1, wherein the image input device is an upconverter arranged to produce output data files of greater pixel count than the input data files.
6. 6. A system according to claim 5, wherein the upconverter interpolates the input data files.
7. 7. A system according to any preceding claim, further comprising a decompressor for expanding the compressed data files.
8. 8. A system according to any preceding claim, further comprising means for arranging de-compressed data into a desired image data file or video format.
9. 9. A system according to any preceding claim, wherein the compression function is JPEG.
10. 10. A method of producing digital image files of comparatively low data size for transmission or storage, comprising: - deriving image data files at a high resolution but with low contrast ranges between adjacent pixels from a source; and - compressing the data files to reduce the size of the data files using a compression function.
11. 11. A method according to claim 10, wherein the step of deriving image data files comprises scanning film with a film scanner at a resolution approximating or exceeding the resolution of the film.
12. 12. A method according to claim 11, wherein the scan resolution is substantially 4K.
13. 13. A method according to claim 11 or 12, wherein the scan resolution is 4096 pixels per line.
14. 14. A method according to claim 10, wherein the step of deriving image data files comprises upconverting an image source from a comparatively low pixel count to a higher pixel count.
15. 15. A method according to claim 14, wherein the step of upconverting comprises interpolating.
16. 16. A method according to any of claims 10 to 14, wherein the compression function is JPEG.
17. 17. A system for producing digital image files from film comprising : - a film scanner for scanning the film at a scanning resolution approximately or exceeding the resolution of the film and producing data files of scanned data; - a compressor for compressing the data files; a decompressor for expanding the compressed data files; and means for arranging the decompressed data into a desired image data file or video format.
18. 18. A system according to claim 17, wherein the scan resolution is 4096 pixels per line.
19. 19. A system according to claim 17 or 18, wherein the compression rate is substantially 52: 1.
20. 20. A system according to claim 18, wherein the scan resolution exceeds the film image resolution.
21. 21. System according to claim 19, wherein the compression ratio can be varied over a wide range to accommodate a quality and file size appropriate to the film image.
22. 22. A system according to any of claims 1 to 9, wherein the compression function is configurable to vary the compression ration over a wide range to accomodate a an appropriate quality and file size.
23. 23. A method according to any of claims 10 to 16, wherein the compression ratio can be varied over a wide range to accommodate an appropriate quality and file size.