Nelson's Video Editing Forum

Since I'm both the Liquid & Edius man & use both software, there are many questions being posted to me regarding the comparison of both products.

So which is better or which software or hardware should I buy?

Based on my own researches & testings, the following post is the finding that I had make. I will have to make it clear that I'm not a programmer or a hardware giz, so the result is purely based on what I see & get. The conclusion is solely based on my own opinion with the main intention of doing an unbiased review on the 2 products.

Please feel free to email me if any of you have any particular addon or feedback to share.

Comparison of the Canopus Edius NX Hardware with Avid Liquid Pro

Output to an External Monitor

The EDIUS NX for HDV provides realtime HD video output. This realtime HD output is

true uncompressed full resolution, full frame rate output. In the case of a 1440 x 1080i

project, the EDIUS hardware is responsible for performing the line scaling from the processed 1440 x 1080i frame to the standard 1920 x 1080i HD frame and sending this video out through the component output of the boards. EDIUS hardware is necessary to perform the same kind line scaling that a HDV camcorder or deck would perform This feature sets the EDIUS solutions apart from L6 claims to perform realtime HDV editing.

Liquid provides cost effective way of viewing HD by offering a full screen preview on the computer screen or HDTV displays through the VGA or DVI output. However, this may lead to several issues :

1. The original component output is in YUV format, which has different dynamic range compared to RGB, with different color representation range. Also many HDTV displays have different gamma characteristics for the YUV input, so input from RGB results in a difference in video quality.

2. In many HDTV displays, RGB and YUV signals have different relationships between the synchronization signal and the video monitoring area. Even if an editor has confirmed the safe area on the HDTV display, when the editing result is written back to the tape and monitored from camera's component output, it may be displayed in a

shifted position.

3. This method does not guarantee synchronization between the audio output from the audio circuit and the video played back from the graphic circuit because there is no way to check the audio and video together from the editing software.

4. The graphics board's RGB output does not have accurate calibration and is not intended for professional video output. It will be difficult to perform accurate color correction by relying on the RGB monitor output from the graphics board.

5. 1080i is an interlaced video signal, however the PC graphics card outputs progressive frames. This causes video clips with dynamic motion to be displayed differently on an RGB monitor from the original HDTV video.

Liquid’s graceful degradation feature allows users to increase the real time playback of the timeline. However, depending on the projects & workflow, there are editors who still prefer the edited video and audio on the monitor to be exactly the same as the video clip which the camera plays back directly from the tape to the HDTV, any dropping of frames or lowering of the video quality is still considered as undesirable. Rhema Broadcasting Group ( RBG ) in Auckland are strong followers of Canopus products. They are believers of Canopus way of defining real time editing, which give users a guarantee of best video quality & no drop frame playback as long as it is within the real time parameter of the hardware as oppose to Liquid’s graceful degradation.

Editing HDV Real-time

Unlike Liquid’s unique Smart Edit technology, Canopus had admitted that editing the native MPEG-2 Transport Stream is complicated, difficult, and impractical. Hence Canopus came out with its own high definition codec. It uses more gentle compression than HDV. It needs less processing power as well, so users can play more simultaneous video streams and have more real-time effects. The main pitfall is the system must be pretty powerful in order to handle the real time conversion of HDV to the HQ codec while capturing.

Basically, The Canopus claimed that their HQ codec is better than HDV and even Sony & DVCPRO-HD codecs. The HQ codec maintains a 4:2:2 color space which is important for keying and compositing , compared to HDV's 4:2:0 sub-sampling. Since the HQ codec is not limited by a specified data rate, the HQ codec uses a variable datarate which can fluctuate according to the need of the video beyond the simple variation provided by a restricted datarate determined by a tape speed. Users can choose to capture HDV in standard or fine mode & customised the quality & file size.

In Canopus’s whitepaper for HQ codec, there is numerous claims that the codec will generate a result that is better the original HDV quality. As the entire process involved a lot of different stages like uncompressing MPEG-2 to another codec, then decompressing this for effects, and finally re-encoding the video to HDV, MPEG-2 (or another format), there is a need of more tests to verify this claims.

Personally, Canopus’s claims had a lot in common with the Cineform Aspect HD’s claims on their website. The company's Aspect HD plug-in for Premiere employs an intermediate codec it calls CFHD, which is based on Wavelet compression. To increase editing performance, the algorithms within the company's Carlsbad software processing engine take advantage of video that's stored in a format that allows more efficient retrieval from hard drives and whose compression format allows for very efficient video effects processing by exploiting Intel SSE2 instructions.

The Aspect HD bundle includes an HDV I/O application for i.LINK transfer. It converts each capture clip from MPEG-2 to CFHD after capture via i.LINK. The new version, now shipping, uses an on-the-fly conversion process. If the system is Pentium 4 2.8GHz or faster, data is converted directly into a file that occupies 33GB ( compare to native HDV at 9 GB per hr ) of space per hour of video. If, however, you have a less powerful Pentium 4, the capture goes into an intelligent buffer that is emptied as fast as the conversion occurs.

Because timecode is not sent from the camcorder while MPEG-2 is captured, like capturing through Edius NX, users cannot set in/out points for capture using the i.LINK HDV I/O application. This prevents batch capture and re-capture. Moreover, video and audio are not presented on your RGB monitor during capture. You can capture multiple shots into a single clip, but whenever a loss of valid data is encountered, capture continues into a new clip.

CineForm emphasizes that CFHD was designed to be a very high-quality intermediate codec operating in a 4:2:2 color space. Quality is high for two reasons. First, because low data rate (19Mbps) HD MPEG-2 is decoded and then recompressed with a high data rate (80Mbps) codec, quality is claimed not to be lost during the capture/conversion process. (SD 480p60 is not yet supported.) Second, because video is only lightly compressed (typically 6:1), during editing it behaves more like uncompressed than compressed video.

For comparing purposes,Canopus NX is installed on a dual Xeon 3.06 system with the Canopus Certified Tyan MOB, 1 GB of RAM, ATI X300-256 RAM, a raid array of 4 x 200 GB HDD. The followings are the test results:

5 min of HDV footage captured at MPEG-2 Transport Stream takes up 960MB
5 min of HDV footage captured at MPEG-2 Program Stream takes up 948MB
5 min of HDV footage captured at standard HQ codec at Q4 & size 30 takes up 2.83GB, almost 3 times the size of the native HDV stream.
Even though Canopus claims that editing in native HDV is possible in Edius NX. During the test, I can only playback 1 stream of native HDV on the timeline. The response is extremely slow when trying to scrub on the timeline, it takes a couple of seconds before the video can be refreshed & previewed on the monitor. The system actually crashed more than 3 times when I was trying to edit both the native HDV transport & program streams. Trying to perform frame accurate scrubbing on the timeline or even editing with native HDV is definitely not a pleasant experience on the NX system.
However, on the same system, I can actually applied 12 filters, in the respective order of :

1-color balance,
2-blur,
3-mosaic,
4-old movie,
5-pencil sketch,
6-raster wipe,
7-tunnel vision,
8-soft focus,
9-mirror,
10-strobe,
11-emboss,
12-anti flicker

on a standard HQ codec clip with real time video out on an external monitor & real time output through firewire without dropping a single frame or any compromise in video quality.

With 1 HQ clip as video background, playing back another HQ clip as 3D PIP ( picture in picture ) is real time. Rendering is only required when more than 4 filters are applied onto the background clip or a 2^Nd 3D PIP is added onto the timeline.
All transitions available in Edius are real time except for those found in Xplode. NX is still able to playback in real time when a real time transition is applied across 2 HQ clips that already had 5 different sets of filters applied on each of them.
Chroma keying of 1 layer of HQ clip over another HQ clip is real time.
I can achieve 18 3D PIPs of DV clips in real time without dropping of a single frame. Based on other research articles on the web, there are users who can achieve up to 20 layers of PIPs when using DV clips.
When dealing with uncompressed clips, I can achieve RT playback with 1 3D PIP or any RT transition across 2 uncompressed clips.

Overall, as long as users are sticking with the HQ codec, the real time playback of NX is fanstatic. Once the users understand the system’s limit & conform to the RT parameters, NX is able to give great assurance & stability to the users. The biggest debate here is which school of thoughts do the users belong to? Liquid’s graceful degradation will give users more & scalable RT capabilities but compromise on the playback quality unless all the effects ( yellow slices ) had been rendered. Canopus’s NX will assure users with full quality playback as long as they are editing within the system’s RT capability.

Generally speaking, users who tend to output the timeline to external devices like VHS or set top DVD recorders preferred the certainty in NX. The idea of having a yellow slice on the Liquid’s timeline which means that there is an uncertainty of the quality of playback ( possibilty of a drop frame ) unless it is rendered still strike fear with these users.

I fullly understand that we can easily overcome this issue by turning the background rendering on all the time but users will normally compare the different systems to see which suits their workflow better. For users who stick to Canopus solutions, the general remarks are “ Why should I render when I don’t even need to do it with Storm or NX”. I am taking this opportunity to raise the feedbacks that I gathered from endusers when I was doing the Liquid roadshows for Pinnacle in Malaysia, Indonesia, Thailand, Vietnam & especially India. The wedding market in these countries were huge, most of the videgraphers were not too concerned with creativity but just straight cuts, transitions & Hollywood FX. In order to cut down the rendering time, the video output is connected straight to a VHS or DVD recorder whereby the video can be recorded as the timeline is being played. Any dropping of frames during this capturing process will means that they will have to redo it again & precious time is being wasted. In this case, Liquid’s graceful degradation & background rendering do not have too much positive appeal to this group of users. Hence if Liquid can give this users an assurance in playback quality by default & the option to turn on graceful degradation if they need more RT preview, I am very confident that Liquid will swing even more users over fom this old school.

The biggest advantage of Edius NX over Liquid Pro will be the EDIUS Hardware which outputs analog component video and audio in HD master quality from a dedicated output circuit. It guarantees realtime, full frame output for realtime playback, and does not drop frames or lower the quality. Audio and video synchronization is complete and guaranteed. The same quality as a broadcast quality HD signal is maintained in the signal level, SN ratio, synchronizing signal level and position, and quality deterioration by horizontal interpolation is prevented. The monitored signal during editing will be of equivalent quality to the content that is written back to the tape and played back from the original playback system.

HDV 1080i (Sony HDV camera format) records in MPEG format with 1440x1080 resolution video to the tape. To accurate edit the video shot, it is necessary to edit in 1440x1080 resolution and monitor in 1920x1080, horizontally stretched through the component output since HDTV displays are designed to receive a 1920x1080i signal. HDTV defines its input signal as an analog signal with its horizontal and vertical display position and size in its resolution. To monitor a high-quality video, a multitap digital filter is required for horizontal interpolation processing. EDIUS Hardware processes it in realtime on the board, matching its display timing with the HDTV requirements.

Another advantage of EDIUS Pro over Liquid will be the pre-process of clips to match the project settings being used in realtime without making any changes to the original clip. So while the SD clip will remain an SD clip on the disk, during playback in an HDV project, there is processing power being used to convert that to the HDV project setting on-the-fly.This pre-processing also comes into play if there is an effect, title, or transition that needs to be applied.

Nelson Ng
Pinnacle Liquid trainer for TVNZ News & Current Affair department
Adobe Premiere certified trainer
Pinnacle Liquid certified trainer
Ex-Lecturer for Singapore Institute of Technology Education
Ex-Producer for Singapore Television Broadcasting Corporation
Product Manager for Avid, Pinnacle, Canopus, Matrox, Ulead, Steinberg & Sony products.

Q1. Where does EDIUS Neo fit into the EDIUS family?
A1. EDIUS Neo is intended to be an entry-level version of the software, omitting the more advanced
editing functions and format support found in the higher level versions. It is important to
remember that the raw realtime editing experience has in no way been compromised or reduced,
as this is a significant factor in attracting users to upgrade to EDIUS Pro. EDIUS Neo is priced
to appeal to the education market, and also open up opportunities for bundling with camera
manufacturers and other OEM vendors.

Q2. So what exactly is the difference between EDIUS Neo software and EDIUS Pro?
A2. A detailed comparison table can be found on page 15 of this Reviewer’s Guide, but the main
differences between EDIUS Neo and EDIUS Pro are the absence of “external” utilities, such as
ProCoder Express for EDIUS, Xplode for EDIUS and TitleMotion Pro for EDIUS; no User Profile
support; simplified Project Settings menus; no audio channel mapping; no advanced tools such as
Multicam and Time Remap; and no support for EDIUS editing hardware, including older boardsets
like the DVStorm series.
One advantage to EDIUS Neo’s freedom from hardware and external utilities, is that it makes the
software 100% compliant with Windows Vista operating systems.

Q3. Does EDIUS Neo require a dongle to function, like the other EDIUS software applications?
A3. No. Since EDIUS Neo is largely aimed at multi-seat environments, such as school campuses, the
requirement for a USB security dongle has been removed.

Q4. Can EDIUS Neo be upgraded to EDIUS Pro?
A4. Yes. There is an upgrade path for users who want to expand their editing feature set at any time.
However, they will need to be aware EDIUS Pro is currently not supported under Windows Vista,
and that EDIUS Pro uses a dongle to function.

Q5. Can EDIUS Neo be upgraded to EDIUS Broadcast software, or any other EDIUS
hardware+software package, such as EDIUS NX Express?
A5. There are currently no upgrade paths to any EDIUS packages other than EDIUS Pro NLE software.