{"text": "这条视频我将手把手教你如何免费无限期使用Codex Cloud Code文本图片视频全部同时实现Token自由。事情是这样的,Agnes AI最近他们官方宣布把自家的文本图片视频模型全部对外开放免费,而且无限期。于是我赶紧用我的Cloud Code接入适用力下,确实是免费的。下面我将手把手教大家如何去使用它,并测试下它真实的模型效果到底是怎么样呢。首先需要在Agnes AI创建API密钥并且复制它。打开CC-Switch这个工具,如果你介入的是Codex,也可以用Codex加加这个工具。在右上角切换到Cloud Code,添加供应商。供应商名称就是Agnes,关我的链接我们填一下。API-K就是我们刚刚复制的API-K,请求地址填写这个。API格式选这个。认证字段我们就选默认的。然后模型映射这儿,我们点击获取模型的列表。我们全部选择Agnes 2.0 Flash。注意这里面是不能选择图像和视频模型的。原因是Cloud Code只支持文本交互,用不了图像和视频模型。解决办法咱们后面再说,我们先把文本模型配置好。点击保存,在列表启用它。再设置本地路由这儿,把这些路由开关全部打开。好了,这样模型就配置好了,现在可以正常使用它了。为了测试它能不能干活,我给了它一个多部组任务。我把写好的一篇稿子的标题发给Cloud Code,并且让它用get笔记的CLA工具找到这篇文章。然后再生成一张指定风格的信息架构图,最后通过飞出发给我。大概过了两到三分钟,结果出来了。我们打开飞出。我们看到它给我们发了一条链接,就是我们刚刚让它生成的这个信息架构图。打开看一下效果。效果,整体效果还算不错,基本符合我们刚才给它的要求。总结一下,文本科兴可以完成工具的调用携带码,指令遵循等多部组任务,也支持图片的士兵,还是可以干活的。但是我们都知道生图和生视频才是套恨销恨的大货。我们看一下直播羊毛,值不值得好。我们前面说过,Cloud Code不能够直接接入图像和视频模型。解决办法就是把图像和视频的API接入文档发给Cloud Code,让它写一个脚本程序去调用。为了以后用起来方便,不用每次都写代码,我把它保存成了两个Scale。下次生图和生视频的时候可以直接调用。我们先来看一下它的文生图能力,用的是Agnus Image 2.1 Flash这个模型。我给它一个产品的idea,让它生成两张效果图看一下。好的,结果出来了,这是第一张,这是第二张。我又给了它一个小故事,让它生成三副的儿童插画绘本。我们来看一下生成的效果。我们可以看到文生图在指定遵循细节描述上都还不错,但这并不算难的。真正考验模型的是给它一张参考图,看它能不能保住主体的一致性。我随手拍了一个杯子,然后让它根据这张照片生成电商产品的效果图。最终的效果基本上符合我给的照片和贴着次要求,但也算不上完美。杯子的颜色和形状还是有一点差别的。那考虑到我这个照片是随手一拍的,如果拍的比较好的情况下,多抽卡几次,可能会还原的更好一些。总结一下图片模型在文本生图上,用来批量产出一些产品矮定的效果,制作绘本插图,文章配图,还是不错的。但有时候也会出现一些令人匪夷所思的情况,比如说这个司马光杂钢。最后我们再来看一下它的视频生成能力,用的是Evnus Video VR.0这个模型。先看一下文生视频,我给了它一段宇航员在红色沙漠星球上行走的提示词。我们看一下效果,效果基本上符合提示词的要求,红色沙漠,陈土飞扬,追踪镜头,这些都有。接着我们再让它用前面生成的智能水杯作为参考图,完成一个宣传视频的制作。视频的场景和运镜基本符合提示词的要求,但是最大的问题就是主体的一致性,这个杯子明显和我给的效果图有差别。还有个小问题,不知道你有没有发现,天盖顶部的温度像计时器一样一直在变,这和实际的物理情况不太相符。所以如果你对视频的一致性,时长要求比较高,拿它制作一些电影漫剧可能比较勾腔。最后我们整体总结一下,Evnus的文本图像视频模型确实是免费的,并且也确实是可以干一些活,但是与顶尖的模型确实有一定的差距。网上有一些反馈说是执行速度慢,我实测下来单张图的生成时间在一分钟左右,单个视频的生成时间在两到三分钟,整体是在我可以接受的file之内,模型的追求效果与提示词的关系是比较大的,最好按照官方的结构栏。如果你有一些常规批量处理的任务,可以耗一波仰谋,确实可以省不少的成本。而今天的视频就到这里,如果视频对你有帮助,别忘了点赞关注,我们下期见。", "segments": [{"id": 0, "seek": 0, "start": 0.0, "end": 29.8, "text": "这条视频我将手把手教你如何免费无限期使用Codex Cloud Code文本图片视频全部同时实现Token自由。事情是这样的,Agnes AI最近他们官方宣布把自家的文本图片视频模型全部对外开放免费,而且无限期。于是我赶紧用我的Cloud Code接入适用力下,确实是免费的。下面我将手把手教大家如何去使用它,并测试下它真实的模型效果到底是怎么样呢。首先需要在Agnes AI创建API密钥并且复制它。", "tokens": [50365, 5562, 48837, 40656, 39752, 1654, 45456, 11389, 16075, 11389, 21936, 2166, 43526, 2347, 235, 18464, 117, 31364, 43446, 16786, 22982, 9254, 34, 1429, 87, 8061, 15549, 17174, 8802, 3919, 122, 16668, 40656, 39752, 38714, 13089, 15729, 24726, 20204, 51, 8406, 45233, 1543, 24675, 1541, 21209, 1546, 11, 37968, 4081, 7318, 46777, 47911, 31929, 9249, 2415, 96, 34688, 16075, 9722, 5155, 1546, 17174, 8802, 3919, 122, 16668, 40656, 39752, 41908, 39823, 38714, 8713, 12022, 18937, 12744, 2347, 235, 18464, 117, 11, 22942, 31364, 43446, 16786, 1543, 37732, 42614, 5266, 114, 7732, 100, 9254, 14200, 32787, 15549, 14468, 14028, 2215, 224, 9254, 13486, 4438, 11, 38114, 106, 24726, 1541, 2347, 235, 18464, 117, 1546, 1543, 47150, 1654, 45456, 11389, 16075, 11389, 21936, 6868, 43526, 6734, 22982, 9254, 11284, 11, 3509, 114, 11038, 233, 5233, 243, 4438, 11284, 6303, 24726, 1546, 41908, 39823, 43076, 9319, 33883, 1541, 48200, 6240, 1543, 36490, 35748, 3581, 37968, 4081, 7318, 2437, 249, 34157, 4715, 40, 47838, 25153, 98, 3509, 114, 20334, 1787, 235, 25491, 11284, 1543, 51855], "temperature": 0, "avg_logprob": -0.08486217676207077, "compression_ratio": 1.35989010989011, "no_speech_prob": 5.925466714518102e-12}, {"id": 1, "seek": 3000, "start": 30.0, "end": 59.32, "text": "打开CC-Switch这个工具,如果你介入的是Codex,也可以用Codex加加这个工具。在右上角切换到Cloud Code,添加供应商。供应商名称就是Agnes,关我的链接我们填一下。API-K就是我们刚刚复制的API-K,请求地址填写这个。API格式选这个。认证字段我们就选默认的。然后模型映射这儿,我们点击获取模型的列表。我们全部选择Agnes 2.0 Flash。注意这里面是不能选择图像和视频模型的。", "tokens": [50365, 12467, 18937, 11717, 12, 38373, 1549, 15368, 23323, 39806, 11, 45669, 30312, 14028, 24620, 34, 1429, 87, 11, 47935, 9254, 34, 1429, 87, 9990, 9990, 15368, 23323, 39806, 1543, 3581, 22060, 5708, 29389, 23632, 26075, 95, 4511, 32787, 15549, 11, 14176, 119, 9990, 3254, 249, 44297, 45581, 1543, 3254, 249, 44297, 45581, 15940, 8204, 108, 5620, 37968, 4081, 11, 28053, 14200, 165, 241, 122, 14468, 15003, 13331, 104, 8861, 1543, 4715, 40, 12, 42, 5620, 15003, 49160, 49160, 1787, 235, 25491, 1546, 4715, 40, 12, 42, 11, 27908, 32718, 10928, 14872, 222, 13331, 104, 5676, 247, 15368, 1543, 4715, 40, 30921, 27584, 2215, 231, 15368, 1543, 7422, 97, 5233, 223, 22381, 28427, 15003, 3111, 2215, 231, 6173, 246, 7422, 97, 1546, 1543, 26636, 41908, 39823, 1431, 254, 1530, 226, 5562, 22933, 11, 15003, 12579, 6336, 119, 31127, 115, 29436, 41908, 39823, 1546, 43338, 17571, 1543, 15003, 38714, 2215, 231, 6852, 102, 37968, 4081, 568, 13, 15, 20232, 1543, 40794, 35102, 8833, 1541, 28590, 2215, 231, 6852, 102, 3919, 122, 12760, 12565, 40656, 39752, 41908, 39823, 1546, 1543, 51831], "temperature": 0, "avg_logprob": -0.14437947352288177, "compression_ratio": 1.3195592286501376, "no_speech_prob": 1.4866838662919157e-11}, {"id": 2, "seek": 5932, "start": 59.32, "end": 69.44, "text": "原因是Cloud Code只支持文本交互,用不了图像和视频模型。解决办法咱们后面再说,我们先把文本模型配置好。点击保存,在列表启用它。", "tokens": [50365, 42577, 1541, 32787, 15549, 14003, 35488, 17174, 8802, 28455, 1369, 240, 11, 9254, 47225, 3919, 122, 12760, 12565, 40656, 39752, 41908, 39823, 1543, 17278, 5676, 111, 39453, 11148, 8975, 109, 9497, 13547, 8833, 8623, 8090, 11, 15003, 10108, 16075, 17174, 8802, 41908, 39823, 38846, 34719, 2131, 1543, 12579, 6336, 119, 24302, 39781, 11, 3581, 43338, 17571, 2392, 107, 9254, 11284, 1543, 50871], "temperature": 0, "avg_logprob": -0.0940816145676833, "compression_ratio": 1.0294117647058822, "no_speech_prob": 1.5442362666373555e-11}, {"id": 3, "seek": 6944, "start": 69.44, "end": 81.48, "text": "再设置本地路由这儿,把这些路由开关全部打开。好了,这样模型就配置好了,现在可以正常使用它了。为了测试它能不能干活,我给了它一个多部组任务。", "tokens": [50365, 8623, 7422, 122, 34719, 8802, 10928, 24658, 23786, 5562, 22933, 11, 16075, 5562, 13824, 24658, 23786, 18937, 28053, 38714, 12467, 18937, 1543, 12621, 11, 21209, 41908, 39823, 3111, 38846, 34719, 12621, 11, 25040, 6723, 15789, 11279, 22982, 9254, 11284, 2289, 1543, 13992, 2289, 11038, 233, 5233, 243, 11284, 8225, 28590, 26111, 25956, 11, 1654, 23197, 2289, 11284, 20182, 6392, 13470, 10115, 226, 26443, 5087, 94, 1543, 50967], "temperature": 0, "avg_logprob": -0.0864687579018729, "compression_ratio": 1.1502890173410405, "no_speech_prob": 1.6161490548616264e-11}, {"id": 4, "seek": 8148, "start": 81.48, "end": 95.16, "text": "我把写好的一篇稿子的标题发给Cloud Code,并且让它用get笔记的CLA工具找到这篇文章。然后再生成一张指定风格的信息架构图,最后通过飞出发给我。大概过了两到三分钟,结果出来了。我们打开飞出。", "tokens": [50365, 1654, 16075, 5676, 247, 20715, 2257, 20878, 229, 10415, 123, 7626, 1546, 162, 3921, 30716, 28926, 23197, 32787, 15549, 11, 3509, 114, 20334, 33650, 11284, 9254, 847, 5437, 242, 34756, 1546, 34, 11435, 23323, 39806, 25085, 4511, 5562, 20878, 229, 17174, 11957, 254, 1543, 26636, 8623, 8244, 11336, 2257, 44059, 25922, 12088, 47209, 30921, 1546, 17665, 26460, 7360, 114, 7360, 226, 3919, 122, 11, 8661, 13547, 19550, 16866, 11808, 252, 7781, 28926, 49076, 1543, 32044, 16866, 2289, 36257, 4511, 10960, 6627, 50064, 11, 45641, 9319, 7781, 34912, 1543, 15003, 12467, 18937, 11808, 252, 7781, 1543, 51049], "temperature": 0, "avg_logprob": -0.08461025507763179, "compression_ratio": 1.106837606837607, "no_speech_prob": 1.4902929584836855e-11}, {"id": 5, "seek": 9516, "start": 95.16, "end": 105.53999999999999, "text": "我们看到它给我们发了一条链接,就是我们刚刚让它生成的这个信息架构图。打开看一下效果。效果,整体效果还算不错,基本符合我们刚才给它的要求。", "tokens": [50365, 15003, 18032, 11284, 23197, 15003, 28926, 2289, 2257, 48837, 165, 241, 122, 14468, 11, 5620, 15003, 49160, 49160, 33650, 11284, 8244, 11336, 1546, 15368, 17665, 26460, 7360, 114, 7360, 226, 3919, 122, 1543, 12467, 18937, 28324, 43076, 9319, 1543, 43076, 9319, 11, 27662, 29485, 43076, 9319, 14852, 19497, 1960, 29900, 11, 37946, 5437, 99, 14245, 15003, 49160, 18888, 23197, 45224, 4275, 32718, 1543, 50884], "temperature": 0, "avg_logprob": -0.08933619755070384, "compression_ratio": 1.3829787234042554, "no_speech_prob": 1.3425288658552859e-11}, {"id": 6, "seek": 9516, "start": 105.75999999999999, "end": 118.4, "text": "总结一下,文本科兴可以完成工具的调用携带码,指令遵循等多部组任务,也支持图片的士兵,还是可以干活的。但是我们都知道生图和生视频才是套恨销恨的大货。我们看一下直播羊毛,值不值得好。", "tokens": [50895, 33440, 45641, 8861, 11, 17174, 8802, 42091, 2347, 112, 6723, 41509, 23323, 39806, 1546, 8897, 225, 9254, 22220, 118, 4845, 99, 23230, 223, 11, 25922, 49061, 3330, 113, 2172, 103, 10187, 6392, 13470, 10115, 226, 26443, 5087, 94, 11, 6404, 35488, 3919, 122, 16668, 1546, 30337, 2347, 113, 11, 45726, 6723, 26111, 25956, 1546, 1543, 11189, 15003, 7182, 7758, 8244, 3919, 122, 12565, 8244, 40656, 39752, 18888, 1541, 1881, 245, 11845, 101, 23049, 222, 11845, 101, 1546, 3582, 18464, 100, 1543, 15003, 28324, 16186, 49993, 7248, 232, 39057, 11, 40242, 1960, 40242, 5916, 2131, 1543, 51527], "temperature": 0, "avg_logprob": -0.08933619755070384, "compression_ratio": 1.3829787234042554, "no_speech_prob": 1.3425288658552859e-11}, {"id": 7, "seek": 11840, "start": 118.4, "end": 130.1, "text": "我们前面说过,Cloud Code不能够直接接入图像和视频模型。解决办法就是把图像和视频的API接入文档发给Cloud Code,让它写一个脚本程序去调用。", "tokens": [50365, 15003, 8945, 8833, 8090, 16866, 11, 32787, 15549, 28590, 1787, 253, 43297, 14468, 14028, 3919, 122, 12760, 12565, 40656, 39752, 41908, 39823, 1543, 17278, 5676, 111, 39453, 11148, 5620, 16075, 3919, 122, 12760, 12565, 40656, 39752, 1546, 4715, 40, 14468, 14028, 17174, 21416, 96, 28926, 23197, 32787, 15549, 11, 33650, 11284, 5676, 247, 20182, 27067, 248, 8802, 29649, 6346, 237, 6734, 8897, 225, 9254, 1543, 50950], "temperature": 0, "avg_logprob": -0.08993820302626666, "compression_ratio": 1.3278236914600552, "no_speech_prob": 1.4615022733699412e-11}, {"id": 8, "seek": 11840, "start": 130.56, "end": 137.72, "text": "为了以后用起来方便,不用每次都写代码,我把它保存成了两个Scale。下次生图和生视频的时候可以直接调用。", "tokens": [50973, 13992, 2289, 3588, 13547, 9254, 9147, 6912, 9249, 27364, 11, 24384, 23664, 9487, 7182, 5676, 247, 19105, 23230, 223, 11, 1654, 42061, 24302, 39781, 11336, 2289, 36257, 7549, 16806, 1220, 1543, 35829, 8244, 3919, 122, 12565, 8244, 40656, 39752, 49873, 6723, 43297, 8897, 225, 9254, 1543, 51331], "temperature": 0, "avg_logprob": -0.08993820302626666, "compression_ratio": 1.3278236914600552, "no_speech_prob": 1.4615022733699412e-11}, {"id": 9, "seek": 11840, "start": 138.08, "end": 142.9, "text": "我们先来看一下它的文生图能力,用的是Agnus Image 2.1 Flash这个模型。", "tokens": [51349, 15003, 10108, 6912, 28324, 45224, 17174, 8244, 3919, 122, 8225, 13486, 11, 9254, 24620, 32, 4568, 301, 29903, 568, 13, 16, 20232, 15368, 41908, 39823, 1543, 51590], "temperature": 0, "avg_logprob": -0.08993820302626666, "compression_ratio": 1.3278236914600552, "no_speech_prob": 1.4615022733699412e-11}, {"id": 10, "seek": 11840, "start": 143.38, "end": 147.02, "text": "我给它一个产品的idea,让它生成两张效果图看一下。", "tokens": [51614, 1654, 23197, 11284, 20182, 1369, 100, 30246, 1546, 482, 64, 11, 33650, 11284, 8244, 11336, 36257, 44059, 43076, 9319, 3919, 122, 28324, 1543, 51796], "temperature": 0, "avg_logprob": -0.08993820302626666, "compression_ratio": 1.3278236914600552, "no_speech_prob": 1.4615022733699412e-11}, {"id": 11, "seek": 14702, "start": 147.02, "end": 150.48000000000002, "text": "好的,结果出来了,这是第一张,这是第二张。", "tokens": [50365, 20715, 11, 45641, 9319, 7781, 34912, 11, 27455, 18049, 44059, 11, 27455, 19693, 44059, 1543, 50538], "temperature": 0, "avg_logprob": -0.07728512315865023, "compression_ratio": 1.3936507936507936, "no_speech_prob": 1.668257372522408e-11}, {"id": 12, "seek": 14702, "start": 151.3, "end": 155.5, "text": "我又给了它一个小故事,让它生成三副的儿童插画绘本。", "tokens": [50579, 1654, 17047, 23197, 2289, 11284, 20182, 7322, 43045, 6973, 11, 33650, 11284, 8244, 11336, 10960, 5935, 107, 1546, 22933, 11957, 98, 11673, 240, 27126, 10115, 246, 8802, 1543, 50789], "temperature": 0, "avg_logprob": -0.07728512315865023, "compression_ratio": 1.3936507936507936, "no_speech_prob": 1.668257372522408e-11}, {"id": 13, "seek": 14702, "start": 156.06, "end": 157.5, "text": "我们来看一下生成的效果。", "tokens": [50817, 15003, 6912, 28324, 8244, 11336, 1546, 43076, 9319, 1543, 50889], "temperature": 0, "avg_logprob": -0.07728512315865023, "compression_ratio": 1.3936507936507936, "no_speech_prob": 1.668257372522408e-11}, {"id": 14, "seek": 14702, "start": 161.4, "end": 166.48000000000002, "text": "我们可以看到文生图在指定遵循细节描述上都还不错,但这并不算难的。", "tokens": [51084, 15003, 6723, 18032, 17174, 8244, 3919, 122, 3581, 25922, 12088, 3330, 113, 2172, 103, 10115, 228, 45161, 11673, 237, 3316, 108, 5708, 7182, 14852, 1960, 29900, 11, 8395, 5562, 3509, 114, 1960, 19497, 46531, 1546, 1543, 51338], "temperature": 0, "avg_logprob": -0.07728512315865023, "compression_ratio": 1.3936507936507936, "no_speech_prob": 1.668257372522408e-11}, {"id": 15, "seek": 14702, "start": 166.48000000000002, "end": 171.38, "text": "真正考验模型的是给它一张参考图,看它能不能保住主体的一致性。", "tokens": [51338, 6303, 15789, 26504, 49657, 234, 41908, 39823, 24620, 23197, 11284, 2257, 44059, 2129, 224, 26504, 3919, 122, 11, 4200, 11284, 8225, 28590, 24302, 21632, 13557, 29485, 1546, 2257, 6784, 112, 21686, 1543, 51583], "temperature": 0, "avg_logprob": -0.07728512315865023, "compression_ratio": 1.3936507936507936, "no_speech_prob": 1.668257372522408e-11}, {"id": 16, "seek": 14702, "start": 171.8, "end": 176.64000000000001, "text": "我随手拍了一个杯子,然后让它根据这张照片生成电商产品的效果图。", "tokens": [51604, 1654, 10673, 237, 11389, 31113, 2289, 20182, 45458, 7626, 11, 26636, 33650, 11284, 31337, 26075, 106, 5562, 44059, 32150, 16668, 8244, 11336, 42182, 45581, 1369, 100, 30246, 1546, 43076, 9319, 3919, 122, 1543, 51846], "temperature": 0, "avg_logprob": -0.07728512315865023, "compression_ratio": 1.3936507936507936, "no_speech_prob": 1.668257372522408e-11}, {"id": 17, "seek": 17702, "start": 177.02, "end": 181.58, "text": "最终的效果基本上符合我给的照片和贴着次要求,但也算不上完美。", "tokens": [50365, 8661, 10115, 230, 1546, 43076, 9319, 37946, 5708, 5437, 99, 14245, 1654, 23197, 1546, 32150, 16668, 12565, 18464, 112, 20708, 9487, 4275, 32718, 11, 8395, 6404, 19497, 1960, 5708, 14128, 9175, 1543, 50593], "temperature": 0, "avg_logprob": -0.1391878032684326, "compression_ratio": 1.191111111111111, "no_speech_prob": 1.2951795885784989e-11}, {"id": 18, "seek": 17702, "start": 181.72, "end": 184.0, "text": "杯子的颜色和形状还是有一点差别的。", "tokens": [50600, 45458, 7626, 1546, 12501, 250, 17673, 12565, 30900, 35276, 114, 45726, 32241, 12579, 21679, 18453, 1546, 1543, 50714], "temperature": 0, "avg_logprob": -0.1391878032684326, "compression_ratio": 1.191111111111111, "no_speech_prob": 1.2951795885784989e-11}, {"id": 19, "seek": 17702, "start": 184.4, "end": 190.94, "text": "那考虑到我这个照片是随手一拍的,如果拍的比较好的情况下,多抽卡几次,可能会还原的更好一些。", "tokens": [50734, 4184, 26504, 12026, 239, 4511, 1654, 15368, 32150, 16668, 1541, 10673, 237, 11389, 2257, 31113, 1546, 11, 13119, 31113, 1546, 11706, 9830, 225, 20715, 46514, 4438, 11, 6392, 46022, 32681, 6336, 254, 9487, 11, 16657, 12949, 14852, 19683, 1546, 19002, 2131, 38515, 1543, 51061], "temperature": 0, "avg_logprob": -0.1391878032684326, "compression_ratio": 1.191111111111111, "no_speech_prob": 1.2951795885784989e-11}, {"id": 20, "seek": 19094, "start": 190.94, "end": 199.18, "text": "总结一下图片模型在文本生图上,用来批量产出一些产品矮定的效果,制作绘本插图,文章配图,还是不错的。", "tokens": [50365, 33440, 45641, 8861, 3919, 122, 16668, 41908, 39823, 3581, 17174, 8802, 8244, 3919, 122, 5708, 11, 9254, 6912, 3416, 117, 26748, 1369, 100, 7781, 38515, 1369, 100, 30246, 5881, 106, 12088, 1546, 43076, 9319, 11, 25491, 11914, 10115, 246, 8802, 11673, 240, 3919, 122, 11, 17174, 11957, 254, 38846, 3919, 122, 11, 45726, 1960, 29900, 1546, 1543, 50777], "temperature": 0, "avg_logprob": -0.16578840240230405, "compression_ratio": 1.158273381294964, "no_speech_prob": 1.3978316767970789e-11}, {"id": 21, "seek": 19094, "start": 199.35999999999999, "end": 203.96, "text": "但有时候也会出现一些令人匪夷所思的情况,比如说这个司马光杂钢。", "tokens": [50786, 8395, 2412, 29111, 6404, 12949, 7781, 20204, 38515, 49061, 4035, 9937, 103, 1787, 115, 5966, 8870, 1546, 46514, 11, 36757, 8090, 15368, 32981, 49970, 20690, 4422, 224, 25153, 95, 1543, 51016], "temperature": 0, "avg_logprob": -0.16578840240230405, "compression_ratio": 1.158273381294964, "no_speech_prob": 1.3978316767970789e-11}, {"id": 22, "seek": 19094, "start": 204.28, "end": 208.7, "text": "最后我们再来看一下它的视频生成能力,用的是Evnus Video VR.0这个模型。", "tokens": [51032, 8661, 13547, 15003, 8623, 6912, 28324, 45224, 40656, 39752, 8244, 11336, 8225, 13486, 11, 9254, 24620, 36, 85, 77, 301, 9777, 13722, 13, 15, 15368, 41908, 39823, 1543, 51253], "temperature": 0, "avg_logprob": -0.16578840240230405, "compression_ratio": 1.158273381294964, "no_speech_prob": 1.3978316767970789e-11}, {"id": 23, "seek": 20870, "start": 208.7, "end": 214.2, "text": "先看一下文生视频,我给了它一段宇航员在红色沙漠星球上行走的提示词。", "tokens": [50365, 10108, 28324, 17174, 8244, 40656, 39752, 11, 1654, 23197, 2289, 11284, 2257, 28427, 2415, 229, 10256, 103, 3606, 246, 3581, 16853, 95, 17673, 3308, 247, 14065, 254, 20682, 28533, 5708, 8082, 9575, 1546, 20949, 25696, 5233, 235, 1543, 50640], "temperature": 0, "avg_logprob": -0.08901297079550254, "compression_ratio": 1.4625, "no_speech_prob": 1.3823928980694067e-11}, {"id": 24, "seek": 20870, "start": 214.61999999999998, "end": 221.1, "text": "我们看一下效果,效果基本上符合提示词的要求,红色沙漠,陈土飞扬,追踪镜头,这些都有。", "tokens": [50661, 15003, 28324, 43076, 9319, 11, 43076, 9319, 37946, 5708, 5437, 99, 14245, 20949, 25696, 5233, 235, 1546, 4275, 32718, 11, 16853, 95, 17673, 3308, 247, 14065, 254, 11, 8842, 230, 45506, 11808, 252, 3416, 105, 11, 39947, 41166, 103, 12373, 250, 39862, 11, 5562, 13824, 48121, 1543, 50985], "temperature": 0, "avg_logprob": -0.08901297079550254, "compression_ratio": 1.4625, "no_speech_prob": 1.3823928980694067e-11}, {"id": 25, "seek": 20870, "start": 221.5, "end": 226.76, "text": "接着我们再让它用前面生成的智能水杯作为参考图,完成一个宣传视频的制作。", "tokens": [51005, 14468, 20708, 15003, 8623, 33650, 11284, 9254, 8945, 8833, 8244, 11336, 1546, 5094, 118, 8225, 15590, 45458, 11914, 13992, 2129, 224, 26504, 3919, 122, 11, 41509, 20182, 2415, 96, 7384, 254, 40656, 39752, 1546, 25491, 11914, 1543, 51268], "temperature": 0, "avg_logprob": -0.08901297079550254, "compression_ratio": 1.4625, "no_speech_prob": 1.3823928980694067e-11}, {"id": 26, "seek": 20870, "start": 227.32, "end": 234.79999999999998, "text": "视频的场景和运镜基本符合提示词的要求,但是最大的问题就是主体的一致性,这个杯子明显和我给的效果图有差别。", "tokens": [51296, 40656, 39752, 1546, 50255, 50218, 12565, 3316, 238, 12373, 250, 37946, 5437, 99, 14245, 20949, 25696, 5233, 235, 1546, 4275, 32718, 11, 11189, 8661, 39156, 34069, 5620, 13557, 29485, 1546, 2257, 6784, 112, 21686, 11, 15368, 45458, 7626, 11100, 1431, 122, 12565, 1654, 23197, 1546, 43076, 9319, 3919, 122, 2412, 21679, 18453, 1543, 51670], "temperature": 0, "avg_logprob": -0.08901297079550254, "compression_ratio": 1.4625, "no_speech_prob": 1.3823928980694067e-11}, {"id": 27, "seek": 23480, "start": 234.8, "end": 241.88000000000002, "text": "还有个小问题,不知道你有没有发现,天盖顶部的温度像计时器一样一直在变,这和实际的物理情况不太相符。", "tokens": [50365, 35091, 7549, 7322, 34069, 11, 17572, 43320, 17944, 28926, 20204, 11, 6135, 5419, 244, 10178, 114, 13470, 1546, 9592, 102, 13127, 12760, 7422, 94, 15729, 34386, 2257, 14496, 34448, 3581, 2129, 246, 11, 5562, 12565, 24726, 8842, 227, 1546, 23516, 13876, 46514, 1960, 9455, 15106, 5437, 99, 1543, 50719], "temperature": 0, "avg_logprob": -0.07602680170977558, "compression_ratio": 1.3686746987951808, "no_speech_prob": 1.3012220641261951e-11}, {"id": 28, "seek": 23480, "start": 241.88000000000002, "end": 247.52, "text": "所以如果你对视频的一致性,时长要求比较高,拿它制作一些电影漫剧可能比较勾腔。", "tokens": [50719, 7239, 45669, 8713, 40656, 39752, 1546, 2257, 6784, 112, 21686, 11, 15729, 32271, 4275, 32718, 11706, 9830, 225, 12979, 11, 24351, 11284, 25491, 11914, 38515, 42182, 16820, 14065, 104, 5935, 100, 16657, 11706, 9830, 225, 7978, 122, 21184, 242, 1543, 51001], "temperature": 0, "avg_logprob": -0.07602680170977558, "compression_ratio": 1.3686746987951808, "no_speech_prob": 1.3012220641261951e-11}, {"id": 29, "seek": 23480, "start": 247.9, "end": 256.78000000000003, "text": "最后我们整体总结一下,Evnus的文本图像视频模型确实是免费的,并且也确实是可以干一些活,但是与顶尖的模型确实有一定的差距。", "tokens": [51020, 8661, 13547, 15003, 27662, 29485, 33440, 45641, 8861, 11, 36, 85, 77, 301, 1546, 17174, 8802, 3919, 122, 12760, 40656, 39752, 41908, 39823, 38114, 106, 24726, 1541, 2347, 235, 18464, 117, 1546, 11, 3509, 114, 20334, 6404, 38114, 106, 24726, 1541, 6723, 26111, 38515, 25956, 11, 11189, 940, 236, 10178, 114, 1530, 244, 1546, 41908, 39823, 38114, 106, 24726, 2412, 24272, 1546, 21679, 6563, 251, 1543, 51464], "temperature": 0, "avg_logprob": -0.07602680170977558, "compression_ratio": 1.3686746987951808, "no_speech_prob": 1.3012220641261951e-11}, {"id": 30, "seek": 23480, "start": 256.86, "end": 263.92, "text": "网上有一些反馈说是执行速度慢,我实测下来单张图的生成时间在一分钟左右,单个视频的生成时间在两到三分钟,", "tokens": [51468, 16469, 239, 5708, 32241, 13824, 22138, 11748, 230, 8090, 1541, 3416, 100, 8082, 31217, 13127, 20645, 11, 1654, 24726, 11038, 233, 4438, 6912, 47446, 44059, 3919, 122, 1546, 8244, 11336, 44848, 3581, 2257, 6627, 50064, 29457, 11, 47446, 7549, 40656, 39752, 1546, 8244, 11336, 44848, 3581, 36257, 4511, 10960, 6627, 50064, 11, 51821], "temperature": 0, "avg_logprob": -0.07602680170977558, "compression_ratio": 1.3686746987951808, "no_speech_prob": 1.3012220641261951e-11}, {"id": 31, "seek": 26392, "start": 263.92, "end": 270.26, "text": "整体是在我可以接受的file之内,模型的追求效果与提示词的关系是比较大的,最好按照官方的结构栏。", "tokens": [50365, 27662, 29485, 1541, 3581, 1654, 6723, 14468, 23151, 1546, 69, 794, 9574, 34742, 11, 41908, 39823, 1546, 39947, 32718, 43076, 9319, 940, 236, 20949, 25696, 5233, 235, 1546, 28053, 25368, 1541, 11706, 9830, 225, 39156, 11, 8661, 2131, 26613, 32150, 31929, 9249, 1546, 45641, 7360, 226, 8987, 237, 1543, 50682], "temperature": 0, "avg_logprob": -0.09584490203857422, "compression_ratio": 1.2388059701492538, "no_speech_prob": 1.4262954532717753e-11}, {"id": 32, "seek": 26392, "start": 270.52000000000004, "end": 275.62, "text": "如果你有一些常规批量处理的任务,可以耗一波仰谋,确实可以省不少的成本。", "tokens": [50695, 45669, 32241, 13824, 11279, 6758, 226, 3416, 117, 26748, 1787, 226, 13876, 1546, 26443, 5087, 94, 11, 6723, 4450, 245, 2257, 30806, 1550, 108, 8897, 233, 11, 38114, 106, 24726, 6723, 2862, 223, 1960, 15686, 1546, 11336, 8802, 1543, 50950], "temperature": 0, "avg_logprob": -0.09584490203857422, "compression_ratio": 1.2388059701492538, "no_speech_prob": 1.4262954532717753e-11}, {"id": 33, "seek": 26392, "start": 275.96000000000004, "end": 280.6, "text": "而今天的视频就到这里,如果视频对你有帮助,别忘了点赞关注,我们下期见。", "tokens": [50967, 11070, 34947, 40656, 39752, 45918, 35102, 11, 13119, 40656, 39752, 8713, 43320, 4845, 106, 37618, 11, 18453, 26677, 2289, 12579, 5266, 252, 28053, 26432, 11, 15003, 4438, 16786, 23813, 1543, 51199], "temperature": 0, "avg_logprob": -0.09584490203857422, "compression_ratio": 1.2388059701492538, "no_speech_prob": 1.4262954532717753e-11}], "language": "zh"}